CN101662288B - Method, device and system for encoding and decoding audios - Google Patents

Abstract

An audio encoding and decoding method, device, and system, the method specifically includes: extracting time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters, and harmonic interval parameters used to characterize an audio signal; parameters, frequency domain envelope parameters, pitch parameters and harmonic interval parameters are encoded and then transmitted to the decoder. The embodiment of the present invention adopts a set of parameters including time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters, and harmonic interval parameters, which reduces the number of parameters required for encoding and reduces the time required for encoding using parameters. The required number of bits can achieve the purpose of encoding the signal with fewer bits.

Description

Audio encoding and decoding method and device, system

technical field

The present invention relates to the technical field of audio coding and decoding, in particular to a method, device and system for parametric audio coding and decoding.

Background technique

An audio signal generally refers to a sound wave with a frequency of 20 Hz to 20 KHz that can be heard by the human ear, and a digital audio signal refers to an audio signal after analog-to-digital conversion. The conversion from analog to digital involves digital sampling at a specified sampling rate and scalar quantization of time-domain discrete signals at a specified resolution.

Audio coding generally refers to the coding method RIRAC (Redundancy and Irrelevancy Removal AudioCoding) that eliminates statistical redundancy and perceptual insensitivity in audio signals, such as transform domain coding. Audio coding can use a lower bit rate to represent the signal, but at the same time coding noise will be introduced into the signal. Using the masking effect of the human auditory system, these noises will be difficult or inaudible after frequency and time domain shaping of the audio signal. Using this audio coding method that eliminates statistical redundancy and perceptual insensitivity in audio signals, higher-quality coding performance can be obtained with a higher number of bits, but when the bandwidth is unstable, the audio quality of this coding method The decline is very noticeable.

Compared with the above-mentioned audio coding methods that eliminate statistical redundancy and perceptual insensitivity in audio signals, using parameters to encode audio is a method that uses concise parameter descriptions to characterize signals. This method can use lower coding rates. Higher coding quality is obtained, wherein the parameters may be a group of parameters including time domain and frequency domain characteristics of the signal. Since such a group of parameters can be represented by a small number of bits, the method of encoding audio using parameters is very suitable for low-rate transmission mechanisms. After transmitting the parameter description to the decoder, the decoder can reconstruct the audio signal according to these parameters. At present, the methods for encoding audio signals using parameters mainly include:

Existing technology one

This method combines the advanced audio coding (Advanced Audio Coding, referred to as AAC), frequency band replication (Spectral Band Replication, referred to as SBR) and parametric stereo (Parametric Stereo, referred to as PS) in the MPEG-4 audio standard. The audio signal is encoded.

Prior art two

This method mainly uses various models, such as harmonic model, transient model, single spectral line model and noise model, to analyze the audio signal, extract the corresponding model parameters, and use these model parameters to restore the audio signal at the synthesis end.

In the process of realizing the present invention, the inventor found that at least the following problems existed in the prior art:

In practical application, in the process of encoding the audio signal in the prior art, it generally needs to use more encoding bits to obtain a larger encoding bandwidth, which puts forward very high requirements on the bandwidth of the transmission channel; In the case of small bandwidth, it will affect the audio quality encoded by this technology.

In the method of prior art 2, it is necessary to use more parameters to describe the signal at the encoding end in order to obtain a higher-quality synthetic audio signal at the decoding end; therefore, the bits that need to be transmitted when this technology is used in practical applications The number is also large; when the transmission capacity of the channel is further reduced, it will affect the audio quality encoded by this technology.

Contents of the invention

Embodiments of the present invention provide an audio encoding and decoding method, device, and system, which can reduce the number of bits required for encoding in applications, thereby realizing encoding signals with fewer bits. At the same time, the embodiment of the present invention also provides a sub-band encoding and decoding processing method and device for an audio signal, which can encode the signal with fewer bits in the process of encoding and decoding the audio signal in the sub-band.

An audio encoding method comprising:

Extracting time domain envelope parameters, frequency domain envelope parameters, pitch parameters and harmonic interval parameters used to characterize the audio signal; encoding the time domain envelope parameters, frequency domain envelope parameters, pitch parameters and harmonic interval parameters After that, it is transmitted to the decoder.

An audio encoding device, comprising:

A parameter extraction unit is used to extract time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters and harmonic interval parameters used to characterize audio signals;

The sending unit is configured to encode the time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters and harmonic interval parameters, and transmit them to the decoding end.

An audio decoding method, comprising:

Decoding the received data to obtain time domain envelope parameters, frequency domain envelope parameters, pitch parameters and harmonic interval parameters used to characterize the audio signal; according to the time domain envelope parameters, frequency domain envelope parameters, The pitch parameter and the harmonic interval parameter synthesize the audio signal.

An audio decoding device, comprising:

The decoding unit is used to decode the received data to obtain time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters and harmonic interval parameters used to characterize the audio signal;

The synthesizing unit is used for synthesizing the audio signal according to the time domain envelope parameter, frequency domain envelope parameter, pitch parameter and harmonic interval parameter.

An audio codec system comprising:

An encoding device, configured to extract time-domain envelope parameters, frequency-domain envelope parameters, tone parameters and harmonic interval parameters used to characterize audio signals; for the time-domain envelope parameters, frequency-domain envelope parameters, tone parameters and After the harmonic interval parameter is encoded, it is sent to the decoding device;

A decoding device, configured to decode the data sent by the encoding device to obtain the time-domain envelope parameter, frequency-domain envelope parameter, tone parameter and harmonic interval parameter; according to the time-domain envelope parameter, frequency-domain envelope parameter, The audio signal is synthesized using domain envelope parameters, pitch parameters, and harmonic interval parameters.

An encoding processing method comprising:

When the audio signal is encoded in the way of frequency division, if the spectral signal of the audio signal in the current frequency band is similar to the spectral signal of the audio signal in the previous frequency band, the time domain envelope parameters and frequency domain parameters used to characterize the audio signal are extracted. Envelope parameters, and encode the time-domain envelope parameters and frequency-domain envelope parameters to send, and at the same time send information indicating that the spectrum signal of the audio signal in the current frequency band is similar to the spectrum signal of the audio signal in the previous frequency band; if the current The spectral signal of the audio signal of the frequency band is not similar to the spectral signal of the audio signal of the previous frequency band, then the time domain envelope parameters, frequency domain envelope parameters, tone parameters and harmonic interval parameters used to characterize the audio signal are extracted, and The time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters and harmonic interval parameters are encoded and sent, and at the same time, information indicating that the spectrum signal of the audio signal in the current frequency band is similar to the spectrum signal of the audio signal in the previous frequency band is sent.

An encoding processing device, comprising:

A judging unit, configured to judge whether the spectral signal of the audio signal in the current frequency band is similar to the spectral signal of the audio signal in the previous frequency band;

The coding unit is used to extract the time-domain envelope parameters used to characterize the audio signal when the spectral signal of the audio signal in the current frequency band is similar to the spectral signal of the audio signal in the previous frequency band according to the judgment result information obtained by the judgment unit and frequency domain envelope parameters; or, when the spectral signal of the audio signal of the current frequency band is not similar to the spectral signal between the audio signal of the previous frequency band, the time domain envelope parameters and the frequency domain envelope used to characterize the audio signal are extracted parameter, pitch parameter and harmonic interval parameter;

a transmission unit, configured to send the similarity information between the spectral signal of the audio signal in the current frequency band obtained by the judging unit and the spectral signal of the audio signal in the previous frequency band, and transmit the time-domain packet of the audio signal extracted by the encoding unit Envelope parameters and frequency domain envelope parameters are encoded and then sent; or, sending the information that the spectrum signal of the audio signal of the current frequency band obtained by the judgment unit is not similar to the spectrum signal between the audio signal of the previous frequency band, and the coded The time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters and harmonic interval parameters of the audio signal extracted by the unit are encoded and then sent.

A decoding processing method, comprising:

Receive the data sent by the encoding end, if the information indicating that the spectral signal of the audio signal in the current frequency band is similar to the spectral signal of the audio signal in the previous frequency band is received, according to the time domain envelope parameters and frequency domain envelope used to characterize the audio signal parametrically synthesized audio signals, wherein the time-domain envelope parameters and frequency-domain envelope parameters are decoded from received data;

If information indicating that the spectral signal of the audio signal of the current frequency band is not similar to the spectral signal of the audio signal of the previous frequency band is received, according to the time domain envelope parameters, frequency domain envelope parameters, tone parameters and Synthesizing an audio signal with harmonic interval parameters, wherein the time domain envelope parameters, frequency domain envelope parameters, pitch parameters and harmonic interval parameters are obtained by decoding from received data.

A decoding processing device is characterized in that it comprises:

The receiving information unit is used to receive information indicating that the spectral signal of the audio signal in the current frequency band is similar to the spectral signal of the audio signal in the previous frequency band, and decode the received data to obtain the time-domain envelope parameters and Frequency-domain envelope parameters; or, receiving information indicating that the spectral signal of the audio signal in the current frequency band is not similar to the spectral signal of the audio signal in the previous frequency band, and decoding the received data to obtain a time-domain packet for representing the audio signal Envelope parameters, frequency domain envelope parameters, pitch parameters and harmonic interval parameters;

a decoding unit, configured to synthesize an audio signal according to the similar information received by the receiving information unit, and the time-domain envelope parameter and the frequency-domain envelope parameter; or, according to the dissimilar information, and the The time domain envelope parameter, the frequency domain envelope parameter, the tone parameter and the harmonic interval parameter are described to synthesize the audio signal.

It can be seen from the technical solutions provided by the above-mentioned embodiments of the present invention that, compared with the parametric audio coding technology based on a certain model in the prior art, the embodiments of the present invention adopt time-domain envelope parameters, frequency-domain envelope parameters, pitch A set of parameters of parameters and harmonic interval parameters reduces the number of parameters required for encoding and reduces the number of bits required for encoding using parameters, which can achieve the purpose of encoding signals with fewer bits .

Description of drawings

FIG. 1 is a schematic flowchart of an audio encoding method provided by an embodiment of the present invention;

FIG. 2 is a schematic flowchart of an audio decoding method provided by an embodiment of the present invention;

FIG. 3 is a schematic flowchart of an encoding processing method according to an embodiment of the present invention;

FIG. 4 is a schematic flowchart of a decoding processing method according to an embodiment of the present invention;

FIG. 5 is a schematic diagram of the processing process at the encoding end in Embodiment 2 of the present invention;

FIG. 6 is a schematic diagram of the processing process at the decoding end in Embodiment 2 of the present invention;

FIG. 7 is a schematic structural diagram of an audio encoding device provided by an embodiment of the present invention;

FIG. 8 is a schematic structural diagram of an audio decoding device provided by an embodiment of the present invention;

FIG. 9 is a schematic structural diagram of an audio codec system provided by an embodiment of the present invention;

Fig. 10 is a schematic structural diagram of an encoding processing device provided by an embodiment of the invention;

FIG. 11 is a schematic structural diagram of a decoding processing device provided by an embodiment of the present invention;

FIG. 12 is a schematic structural diagram of a decoding unit provided by an embodiment of the present invention.

Detailed ways

In order to obtain a larger encoding bandwidth with a lower encoding rate and obtain higher encoding quality on the basis of existing audio encoding, an embodiment of the present invention provides an audio encoding method, which can specifically extract time Domain envelope parameters, frequency domain envelope parameters, pitch parameters and harmonic interval parameters; after encoding the time domain envelope parameters, frequency domain envelope parameters, pitch parameters and harmonic interval parameters, transmit them to the decoding end.

Further, when the harmonic interval of the audio signal is different from the value of the first harmonic offset, extract the first harmonic offset parameter of the audio signal, encode it and transmit it to the decoder end.

FIG. 1 is a schematic flowchart of an audio coding method according to an embodiment of the present invention. The audio coding method according to this embodiment of the present invention will be described below in conjunction with FIG. 1 . As shown in Figure 1, the details may include:

11: Extract the time-domain envelope parameters of the audio signal that needs to be encoded; specifically, the time-domain envelope of the signal can be obtained by calculating the sub-frame energy of the audio signal, or the signal can be transformed into the frequency domain (or transform domain ) to extract the autoregressive (AR, Auto Regressive) model parameters to characterize the time domain envelope of the signal;

12: Extract the frequency domain envelope parameters of the audio signal; specifically, the frequency domain envelope of the signal can be obtained by calculating the subband energy in the frequency domain (or transform domain), or the white regression model parameters of the signal can be extracted in the time domain To characterize the frequency domain envelope of the signal;

13: Extract the tone parameter of the audio signal; the tone parameter represents the ratio between the harmonic signal and the noise signal in the audio signal; there are many ways to express the tone parameter, which can be the ratio of the maximum value to the minimum value of the autocorrelation function;

14: Extract the harmonic interval (PG, Pitch Grid) parameter of the audio signal; the harmonic interval parameter characterizes the interval between different harmonics of the signal; specifically, the harmonic interval parameter can be estimated by the peak extraction method;

15: Extract the first harmonic offset parameter (P0, Pitch Offset); specifically, the first harmonic offset parameter can be estimated according to the harmonic interval parameter, and the first harmonic offset parameter can be encoded Transmission; the first harmonic offset parameter represents the position of the first harmonic of the audio signal; it should be pointed out that if the value of the first harmonic offset is equal to the harmonic interval, this step can be omitted; that is When the harmonic interval of the audio signal is different from the value of the first harmonic offset, extracting the first harmonic offset parameter of the audio signal;

After the above-mentioned time-domain envelope parameter, frequency-domain envelope parameter, pitch parameter, harmonic interval parameter and first harmonic offset parameter are coded (or quantized and then coded), they are output.

It should be pointed out that the above pitch parameter, harmonic interval parameter and first harmonic offset parameter can be calculated in frequency domain (or transform domain), for example, can also be calculated in time domain. Moreover, the order of obtaining the above parameters is not unique, that is, no matter in what order, as long as the time domain envelope parameters, frequency domain envelope parameters, pitch parameters, harmonic interval parameters and first harmonic offset of the above audio signal are obtained Quantitative parameters are enough.

The above content describes the audio coding method flow of the embodiment of the present invention. Through the above method, the parameters including time domain envelope parameters, frequency domain envelope parameters, pitch parameters, harmonic interval parameters and first harmonic offset parameters can be used. A set of parameters, or a set of parameters including time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters and harmonic interval parameters, to characterize an audio signal. Compared with the parametric audio coding technology based on a certain model in the prior art, the set of parameters adopted in the embodiment of the present invention reduces the number of parameters required for coding, and simultaneously reduces the number of bits required for coding using parameters; Thereby solving the problem that the number of bits of the traditional RIRAC coding method is higher; Simultaneously, compared with the existing parametric audio coding algorithm, because this group of parameters of the embodiment of the present invention can be encoded with less number of bits, thereby further reducing The encoding rate of the signal, and when the transmission capacity of the channel is constant, because the number of encoding bits of the present invention is low, it can encode a signal with a higher bandwidth, and realizes obtaining a larger encoding bandwidth and a higher encoding bandwidth with a lower encoding rate. High encoding quality.

The embodiment of the present invention also provides an audio decoding method, which may specifically include: decoding the received data to obtain time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters and harmonic intervals used to characterize the audio signal Parameters; synthesize an audio signal according to the time domain envelope parameters, frequency domain envelope parameters, pitch parameters and harmonic interval parameters.

Further, the method further includes: decoding the received data including the first harmonic offset parameter to obtain the first harmonic offset parameter used to characterize the audio signal.

The step of synthesizing audio signal comprises:

Obtain a harmonic signal according to the harmonic interval parameter; or when the harmonic interval of the audio signal is different from the first harmonic offset parameter, according to the harmonic interval parameter and the first harmonic offset Quantitative parameters, get the harmonic signal;

adjusting the ratio between the harmonic signal and the noise signal according to the tone parameter; and obtaining a reconstructed spectrum signal according to the adjusted harmonic signal and noise signal;

Processing the reconstructed spectrum signal according to the frequency domain envelope parameter and the time domain envelope parameter to obtain a synthesized audio signal.

FIG. 2 is a schematic flowchart of an audio decoding method provided by an embodiment of the present invention. The audio decoding method of the embodiment of the present invention will be introduced below in conjunction with FIG. 2 . As shown in Figure 2, the details may include:

21: Decode the received data to obtain the time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters and harmonic interval parameters used to characterize the audio signal. When the harmonic interval of the audio signal is offset from the first harmonic When the value of the displacement is different, the first harmonic offset parameter is also included;

22: Obtain the harmonic signal according to the harmonic interval parameter (when the harmonic interval of the audio signal is different from the first harmonic offset parameter, according to the harmonic interval parameter and the first harmonic offset parameter, obtain harmonic signal; otherwise the value of the first harmonic offset is equal to the value of the harmonic interval); this harmonic structure can be represented by harmonics with random phases, where the first harmonic offset parameter determines the first The position of the harmonic, the interval of each harmonic is determined by the harmonic interval parameter; the harmonic structure is the harmonic signal;

23: Generate a noise signal, for example, a noise signal can be generated by a random number generator;

24: Adjust the ratio between the harmonic signal and the noise signal according to the value of the tone parameter; and obtain a reconstructed spectrum signal according to the adjusted harmonic signal and noise signal;

25: Perform frequency domain shaping processing on the reconstructed spectral signal according to the frequency domain envelope parameters to obtain a frequency domain shaped signal; for example, denormalize the reconstructed spectral signal according to the decoded subband energy envelope After processing, the signal after frequency domain shaping is obtained;

26: Perform time-domain shaping processing on the frequency-domain shaped signal according to the time-domain envelope parameters to obtain a final synthesized audio signal; for example, the frequency-domain shaped signal may be transformed according to the decoded subframe energy envelope After going to the time domain and then performing denormalization processing, the final synthesized audio signal is obtained.

It should be pointed out that the order of frequency-domain shaping and time-domain shaping is not unique, and the reconstructed spectrum signal can also be processed in time-domain according to the time-domain envelope parameters first, and then the shaped spectrum can be processed according to the frequency-domain envelope parameters. The signal is subjected to frequency-domain shaping processing to obtain the final synthesized audio signal.

The above content describes the audio decoding method flow of the embodiment of the present invention. The embodiment of the present invention provides time-domain envelope parameters, frequency-domain envelope parameters, tone parameters, harmonic interval parameters and first A set of parameters of the harmonic offset parameters can realize the synthesis of audio signals with fewer bits, and the audio signal quality is higher; and, when the harmonic structure of the audio signal is obvious, the audio quality obtained by decoding is better good.

To facilitate the understanding of the embodiments of the present invention, the specific implementation solutions of encoding and decoding of the embodiments of the present invention will be described in detail below.

Embodiment one

In this embodiment, the encoder extracts the time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters and harmonic interval parameters of the audio signal respectively. Since the harmonic interval of the audio signal and the first harmonic deviation The offset parameters are the same, so the step of extracting the first harmonic offset parameter is omitted; after receiving the above parameters, the decoder performs decoding according to the above parameters to obtain a synthesized audio signal.

The implementation process at the encoding end may specifically include:

(1): Extract the time-domain envelope parameters of the signal: For example, by calculating the sub-frame energy of the audio signal to obtain the time-domain envelope parameters of the signal, the sub-frame energy envelope {temp_env(0), temp( 1),..., temp(N-1)}, where N is the number of subframes, if the frame length is 15ms, and the subframe length is 3ms, then N=5; the energy envelope of this subframe is quantized, namely The time-domain envelope parameter is obtained, and the time-domain envelope parameter can be further encoded; at the same time, the quantized time-domain envelope can be used to perform time-domain normalization processing on the signal;

Of course, in practical applications, the signal can also be transformed into the frequency domain (or transform domain) and then the autoregressive (AR, AutoRegressive) model parameters can be extracted to characterize the time domain envelope of the signal;

(2): Extract the frequency domain envelope parameters of the audio signal; for example, when the autoregressive model parameters of the signal are extracted in the time domain to characterize the frequency domain envelope of the signal, the autoregressive model parameters {α ₀ of the signal are calculated in the time domain _{. _} , get the residual signal err(n);

In specific applications, the frequency domain envelope parameters of the signal can also be obtained by calculating the subband energy in the frequency domain (or transform domain);

(3): Extract the tone parameter of the audio signal; the tone parameter represents the ratio between the harmonic signal and the noise signal in the audio signal; there are many ways to express the tone parameter, which can be between the maximum value and the minimum value of the autocorrelation function The ratio, such as T=max(ACF(k ₀ ))/min(ACF(k ₀ )), can also be in other forms, as long as the proportional relationship between harmonic and noise can be represented; among them, the autocorrelation function The calculation of ACF(k ₀ ) can be obtained by FFT transformation and inverse FFT transformation. For example, FFT transformation is performed on the residual signal err(n) in (2), and the frequency domain signal S(k)=FFT(err(n )), and further obtain the autocorrelation function ACF(k ₀ )=IFFT(|FFT(S(k))| ² ); of course, it can also be calculated directly, for example $\mathrm{ACF}\left(k_0\right)=\underset{k=0}{\overset{L-1}{\mathrm{\Σ}}}S\left(k\right)S(k+k_0),$ Where L is the number of frequency-domain transform coefficients within the encoding bandwidth; in addition, the average amplitude difference function (AMDF, Average Mean Difference Function) can also be used to correct the autocorrelation function;

(4): Extract the harmonic interval (PG, Pitch Grid) parameter of the audio signal; the harmonic interval parameter characterizes the interval between different harmonics of the signal; specifically, the integer part of the harmonic interval parameter can be estimated by the peak extraction method , for example, the integer part of the harmonic interval parameter can be obtained by calculating PG=arg max(ACF(k ₀ )); the fractional value of the harmonic interval can be obtained by the method of peak extraction after interpolating the autocorrelation function ACF(k ₀ ); specifically Therefore, the interpolation calculation of the autocorrelation function can be performed only near the integer harmonic interval obtained first, and the fractional value of the harmonic interval can be searched in the interpolated autocorrelation function; in order to obtain better performance, the The obtained harmonic interval parameters are further modified and then encoded and transmitted to suppress the generation of multiplied and fractional frequencies; for example, compare the obtained harmonic interval PG of the current frame with the harmonic interval old_PG of the previous frame, if The ratio between the harmonic interval of the current frame and the harmonic interval of the previous frame is less than a certain threshold value (such as 0.1) and ACF(old_PG)>0.95ACF(PG), then the harmonic interval of the previous frame is used instead of this frame The obtained harmonic interval PG=old_PG;

(5): Since the value of the first harmonic offset in this embodiment is equal to the harmonic interval, this step can be omitted; but when the value of the first harmonic offset is not equal to the harmonic interval, extract the first harmonic Specifically, when using the harmonic offset parameter, the first harmonic offset parameter can be estimated according to the harmonic interval parameter, and the first harmonic offset parameter is encoded and transmitted; the first harmonic offset parameter represents the audio The position of the first harmonic of the signal; it should be pointed out that if the value of the first harmonic offset is equal to the harmonic interval, then this step can be omitted; that is, when the harmonic interval of the audio signal is the same as the first harmonic When the values of the wave offset are different, the first harmonic offset parameter of the audio signal is extracted;

Encode (or output after quantization) the above time domain envelope parameters, frequency domain envelope parameters, pitch parameters and harmonic interval parameters. Of course, if (5) is not omitted, the first harmonic offset parameter will also be encoded and transmitted.

It should be pointed out that the above pitch parameter, harmonic interval parameter and first harmonic offset parameter can be calculated in frequency domain (or transform domain), for example, can also be calculated in time domain. Moreover, the order of obtaining the above parameters is not unique, that is, no matter in what order, as long as the time domain envelope parameters, frequency domain envelope parameters, pitch parameters, harmonic interval parameters and first harmonic offset of the above audio signal are obtained Quantitative parameters can be;

Correspondingly, the decoding end decodes the received data to obtain time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters and harmonic interval parameters used to characterize the audio signal, and synthesize the audio signal. Of course, if the encoding end (5) is not omitted, the parameters decoded by the decoding end also include the first harmonic offset parameter.

The specific process of decoding at the decoding end may include:

(6): Decode the received data to obtain the time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters and harmonic interval parameters used to characterize the audio signal; of course, if the harmonic interval of the audio signal at the encoding end When it is different from the value of the first harmonic offset, the first harmonic offset parameter is also obtained;

(7): The harmonic signal is obtained according to the harmonic interval parameter; the harmonic structure can be represented by harmonics with random phases, where the position of the first harmonic is equal to the value of the harmonic interval, and the interval of each harmonic is also given by The harmonic interval parameter is determined; the harmonic structure is the harmonic signal; specifically, for example: starting from the starting frequency point according to the harmonic interval expressed by the harmonic interval parameter (PG), the harmonics with random phases will be pulsed The form is placed at the corresponding frequency point within the signal bandwidth range, thereby generating a harmonic signal buf_pulses(k), for example $\mathrm{buf}\_\mathrm{pulses}\left(k\right)=h\left(k\right)*\underset{\underset{0\≤\mathrm{no}*\mathrm{PG}\≤L-1}{\mathrm{no}\∈Z}}{\mathrm{\Σ}}\mathrm{\δ}(k-(\mathrm{no}*\mathrm{PG})),$ where h(k) denotes a harmonic with random phase;

It should be noted that if the decoding end also receives the first harmonic offset parameter, the decoding end can obtain the harmonic signal according to the harmonic interval parameter and the first harmonic offset parameter; the harmonic structure can be Represented by harmonics with random phases, where the First Harmonic Offset parameter determines the position of the first harmonic, and the spacing of individual harmonics is determined by the Harmonic Interval parameter; this harmonic structure is the harmonic signal. Specifically, for example, the first harmonic offset parameter (P0) is the position of the first pulse, and the harmonic interval represented by the harmonic interval parameter (PG) from the first pulse position will have a random phase The harmonics of are placed in the corresponding frequency points within the signal bandwidth in the form of pulses, thereby generating the harmonic signal buf_pulses(k), for example $\mathrm{buf}\_\mathrm{pulses}\left(k\right)=h\left(k\right)*\underset{\underset{0\≤\mathrm{PO}+\mathrm{no}*\mathrm{PG}\≤L-1}{\mathrm{no}\∈Z}}{\mathrm{\Σ}}\mathrm{\δ}(k-(\mathrm{PO}+\mathrm{no}*\mathrm{PG})),$ where h(k) denotes a harmonic with random phase;

(8): Generate noise signal, for example, noise signal buf_noise(k) can be generated by a random number generator;

(9): adjust the ratio between the harmonic signal and the noise signal according to the value of the tone parameter; and obtain the reconstructed spectral signal according to the adjusted harmonic signal and noise signal; there are many kinds of specific adjustments, for example: first Calculate the energy of the harmonic signal and the noise signal respectively, denoted as enerP and enerN, and then calculate the adjustment factor β ₁ =1-T and ${\mathrm{\β}}_2=\sqrt{\frac{\mathrm{enerP}}{\mathrm{ener\; N}}}*T,$ Where T is the tone parameter; and get the reconstructed spectral signal after correction $\hat{S}\left(k\right)={\mathrm{\β}}_1\mathrm{buf}\_\mathrm{pulses}\left(k\right)+{\mathrm{\β}}_2\mathrm{buf}\_\mathrm{noise}\left(k\right);$ Transform the reconstructed spectral signal to the time domain by inverse FFT transformation, denoted as

进行逆滤波，得到频域整形后的信号

(10): Perform frequency-domain shaping processing on the reconstructed spectrum signal according to the frequency-domain envelope parameters to obtain a signal after frequency-domain shaping; for example, according to the autoregressive model parameters obtained by decoding, the signal is

Perform inverse filtering to obtain the signal after frequency domain shaping

进行去归一化处理后，得到最终的合成音频信号。(11): Perform time-domain shaping processing on the frequency-domain shaped signal according to the time-domain envelope parameters to obtain the final synthesized audio signal; for example, the signal can be processed according to the decoded sub-frame energy envelope

After denormalization processing, the final synthesized audio signal is obtained.

Compared with the parametric audio coding technology based on a certain model in the prior art, the set of parameters adopted in the embodiment of the present invention reduces the number of parameters required for coding, and simultaneously reduces the number of bits required for coding using parameters; Thereby solving the problem that the number of bits of the existing coding method is higher; at the same time, compared with the existing parametric audio coding algorithm, because this group of parameters in the embodiment of the present invention can be coded with less number of bits, thereby further reducing the signal coding rate, and when the transmission capacity of the channel is constant, because the number of coding bits of the present invention is low, it can code a signal with a higher bandwidth, and achieve a larger coding bandwidth and a higher coding rate with a lower coding rate. encoding quality. At the same time, the audio signal can be synthesized with fewer bits at the decoding end, and the quality of the audio signal is high; and, when the harmonic structure of the audio signal is obvious, the audio quality obtained through decoding is better.

An embodiment of the present invention also provides an encoding processing method, which may specifically include: when encoding an audio signal by frequency division, if the spectral signal of the audio signal in the current frequency band is similar to the spectral signal of the audio signal in the previous frequency band , then extract the time-domain envelope parameters and frequency-domain envelope parameters used to characterize the audio signal, encode the time-domain envelope parameters and frequency-domain envelope parameters and send them, and at the same time send the spectrum representing the audio signal in the current frequency band Information that the signal is similar to the spectral signal of the audio signal of the previous frequency band; if the spectral signal of the audio signal of the current frequency band is not similar to the spectral signal of the audio signal of the previous frequency band, extract the time domain envelope parameters used to characterize the audio signal , frequency domain envelope parameters, tone parameters and harmonic interval parameters, and encode the time domain envelope parameters, frequency domain envelope parameters, tone parameters and harmonic interval parameters, and send an audio signal representing the current frequency band at the same time Information that the spectral signal of the audio signal is similar to that of the audio signal of the previous frequency band.

Specifically, the information indicating that the spectral signal of the audio signal in the current frequency band is similar or dissimilar to the spectral signal of the audio signal in the previous frequency band can be specifically represented by a coding mode parameter; the coding mode parameter is used to indicate that the decoding end When the spectral signal of the audio signal of the current frequency band is similar to the spectral signal of the audio signal of the previous frequency band, decoding the audio signal of the current frequency band according to the time domain envelope parameter and the frequency domain envelope parameter of the audio signal; or Indicates that when the spectral signal of the audio signal in the current frequency band is not similar to the spectral signal of the audio signal in the previous frequency band, according to the time domain envelope parameter, frequency domain envelope parameter, pitch parameter and harmonic interval of the audio signal Parameters to decode the audio signal in the current frequency band.

Further, if the spectral signal of the audio signal in the current frequency band is not similar to the spectral signal of the audio signal in the previous frequency band, and when the harmonic interval of the audio signal is different from the value of the first harmonic offset, the extracted The first harmonic offset parameter of the audio signal; and transmit the first harmonic offset parameter to the decoding end. Moreover, if the spectral signal of the audio signal in the current frequency band is similar to the spectral signal of the audio signal in the previous frequency band, the tone parameters of the audio signal may also be extracted and transmitted to the decoding end.

Correspondingly, the embodiment of the present invention also provides a decoding processing method, which may specifically include: receiving the data sent by the encoding end, if the received spectral signal representing the audio signal of the current frequency band is similar to the spectral signal of the audio signal of the previous frequency band information, synthesizing the audio signal according to the time-domain envelope parameters and frequency-domain envelope parameters used to characterize the audio signal, wherein the time-domain envelope parameters and the frequency-domain envelope parameters are obtained by decoding from the received data; If information indicating that the spectral signal of the audio signal of the current frequency band is not similar to the spectral signal of the audio signal of the previous frequency band is received, according to the time domain envelope parameters, frequency domain envelope parameters, tone parameters and Synthesizing an audio signal with harmonic interval parameters, wherein the time domain envelope parameters, frequency domain envelope parameters, pitch parameters and harmonic interval parameters are obtained by decoding from received data.

Specifically, according to the received encoding mode parameter, it is determined whether the spectral signal of the audio signal of the current frequency band is similar or dissimilar to the spectral signal of the audio signal of the previous frequency band; if the spectral signal of the audio signal of the current frequency band is similar to the previous The spectral signals between the audio signals of the frequency bands are similar, then according to the received time domain envelope parameters and frequency domain envelope parameters used to characterize the audio signals, the audio signals are synthesized; If the spectral signals of the audio signals in the frequency bands are not similar, the audio signals are synthesized according to the received time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters and harmonic interval parameters.

If the spectral signal of the audio signal of the current frequency band is not similar to the spectral signal of the audio signal of the previous frequency band, the received time domain envelope parameter, frequency domain envelope parameter, tone parameter and harmonic interval parameter may also include : The first harmonic offset parameter of the audio signal; if the spectral signal of the audio signal of the current frequency band is similar to the spectral signal of the audio signal of the previous frequency band, the received time domain used to characterize the audio signal The envelope parameter and the frequency domain envelope parameter may also include: a tone parameter used to characterize the audio signal.

FIG. 3 is a schematic flowchart of an encoding processing method according to an embodiment of the present invention, and the encoding processing method according to this embodiment of the present invention will be introduced below with reference to FIG. 3 . As shown in Figure 3, the details may include:

31: When the audio signal is encoded by frequency division, judge whether the spectral signal of the audio signal in the current frequency band is similar to the spectral signal of the audio signal in the previous frequency band; specifically, it can be expressed by determining the encoding mode parameter CM ; For example, the cross-correlation between the current frequency band signal spectrum and the previous frequency band signal spectrum can be calculated first to determine the similarity between the current frequency band harmonic structure and the previous frequency band harmonic structure; when the cross-correlation is greater than a certain threshold value, it can be determined that the current frequency band harmonic structure is similar to the previous frequency band harmonic structure, and CM is set to 1, otherwise CM is set to 0; and the current frequency band signal spectrum and the previous frequency band signal spectrum When similar, the following pitch parameters, harmonic interval parameters and first harmonic offset parameters may no longer be extracted;

32: If similar, extract the time domain envelope parameters and frequency domain envelope parameters used to characterize the audio signal; if not, extract the time domain envelope parameters, frequency domain envelope parameters, and pitch used to characterize the audio signal parameters and harmonic interval parameters; that is, if the current frequency band signal spectrum is similar to the previous frequency band signal spectrum, the pitch parameter, harmonic interval parameter and first harmonic offset parameter of the audio signal may not be extracted; Specifically, the method for extracting the above parameters may be as follows:

Extract time-domain envelope parameters; for example, by calculating the subframe energy envelope and global gain factor gain of the current frequency band signal, and judging whether the signal is a steady-state signal or a transient signal according to these two values; if it is a steady-state signal, then Quantize the global gain factor gain, and use the obtained quantization value as the time domain envelope parameter; if it is a transient signal, quantize the subframe energy envelope, and use the obtained quantization value as the time domain envelope parameter; and according to The time domain envelope parameter performs time domain normalization processing on the current frequency band signal to obtain the time domain normalized signal;

Extract frequency domain envelope parameters; for example, after MDCT (modified discrete cosine transform, Modified Discrete Cosine Transform) transformation is performed on the signal after time domain normalization, a set of MDCT coefficients is obtained, which corresponds to the frequency band after instant domain normalization For frequency domain signals, when processing the frequency domain signals, divide the group of frequency domain signals into N subbands, extract and quantize the sub-generation energy of each subband, and obtain a set of quantized frequency domain envelopes, which is the frequency domain envelope Envelope parameter; Carry out frequency domain normalization processing to frequency domain signal according to frequency domain envelope parameter, obtain the signal after frequency domain normalization;

Extract pitch parameters; specifically, parameter extraction can be performed directly in the MDCT domain; in order to further improve the performance of the encoder, it is also possible not to directly perform parameter extraction in the MDCT domain, but to calculate the pseudo-spectrum signal according to the original frequency domain signal, and according to this Pseudospectral signal calculation pitch parameters; pitch parameters can be represented by the ratio between the maximum value and the minimum value of the autocorrelation function, where the maximum and minimum values are obtained in the expected range or are beneficial to the calculation of harmonic interval parameters carried out within the scope of

Extract the harmonic interval parameter PG; the harmonic interval parameter of the high-frequency band signal is usually extracted in the frequency domain (or transform domain); the integer value of the harmonic interval can be estimated by the autocorrelation function by the peak extraction method, and the harmonic interval The fractional value of the wave interval can be estimated from the interpolated autocorrelation function by the method of peak extraction; the interpolation calculation of the autocorrelation function can also be performed only near the obtained integer harmonic interval, and then the harmonic can be obtained by the method of peak extraction fractional value of the wave interval;

Extract the first harmonic offset parameter, for example, estimate the first harmonic offset parameter P0 according to the harmonic interval; specifically, within the range of the harmonic interval, that is, within the range of [0, PG], the first harmonic The wave components are placed at different offset positions, and other harmonics are placed in sequence according to the harmonic interval, and the correlation between the resulting spectrum and the pseudo spectrum is calculated. The offset position with the greatest correlation is the first Harmonic offset; at the same time, the first harmonic offset parameter can also be used to further modify the estimated value of the harmonic interval parameter, so as to achieve a better parameter extraction effect; it should be pointed out that if the first harmonic offset If the value of the displacement is always equal to the harmonic interval, this step can be omitted;

33: Send information indicating that the spectral signal of the audio signal in the current frequency band is similar or dissimilar to the spectral signal of the audio signal in the previous frequency band; for example, encode the encoding mode parameters and send them; and send the extracted parameters after encoding; specifically , when CM is equal to 1, a set of parameters including encoding mode parameters, time domain envelope parameters and frequency domain envelope parameters will be quantized or encoded, and transmitted to the decoder; when CM is equal to 0, encoding mode is included A set of parameters, including time domain envelope parameters, frequency domain envelope parameters, pitch parameters and harmonic interval parameters, will be quantized, coded, and transmitted to the decoder;

It should be pointed out that when CM is equal to 1, the parameters transmitted to the decoding end can also include tone parameters; when CM is equal to 0, if the value of the first harmonic offset is not equal to the harmonic interval, the first harmonic must be transmitted Wave offset parameter.

Correspondingly, the decoder receives a set of parameters including the encoding mode parameters, time domain envelope parameters, and frequency domain envelope parameters, or receives the above-mentioned parameters including encoding mode parameters, time domain envelope parameters, and frequency domain envelope parameters. A set of parameters, a pitch parameter, and a harmonic interval parameter, synthesizes an audio signal.

It should be pointed out that if the encoding end also transmits the tone parameter when CM is equal to 1, the corresponding decoding end also needs to receive the tone parameter; if the encoding end also transmits the first harmonic offset parameter when CM is equal to 0, the corresponding The decoding end of also needs to receive the first harmonic offset parameter.

Fig. 4 is a schematic flow chart of the decoding processing method according to an embodiment of the present invention;

41: Receive information indicating that the spectral signal of the audio signal in the current frequency band is similar to the spectral signal of the audio signal in the previous frequency band, or information that is not similar; for example, according to the received data, the encoding mode parameter CM is decoded, according to the The encoding mode parameter CM can determine whether they are similar;

42: When the spectral signal of the audio signal in the current frequency band is similar to the spectral signal of the audio signal in the previous frequency band, according to the time-domain envelope parameters and frequency-domain envelope parameters used to characterize the audio signal obtained by decoding the received data, Synthesizing the audio signal; when not similar, synthesizing the audio signal according to the time domain envelope parameters, frequency domain envelope parameters, pitch parameters and harmonic interval parameters used to characterize the audio signal obtained by decoding the received data;

Specifically, when reconstructing the spectral signal:

If the signal spectrum of the current frequency band is similar to the signal spectrum of the previous frequency band, for example, CM is equal to 1, the spectral signal of the previous frequency band can be used as the spectral signal of the current frequency band reconstruction in the way of spectrum replication; of course, different spectrum replication can also be used Reconstruct the spectral signal in the same way; if the encoding terminal CM is equal to 1, the pitch parameter is also transmitted, and the pitch parameter can also be decoded from the code stream, and the spectral signal of the current frequency band is reconstructed through the spectrum of the previous frequency band by means of spectrum copying; specifically, According to the tone parameter, the spectral signal of the previous frequency band can be shaped to obtain the reconstructed spectral signal after shaping, and the shaped spectral signal can be used as the reconstructed spectral signal of the current frequency band;

If there is no correlation between the signal spectrum of the current frequency band and the signal spectrum of the previous frequency band, for example, CM is equal to 0, the tone parameter, harmonic interval parameter and first harmonic offset parameter are decoded from the code stream, and according to the harmonic Harmonic signals are obtained from the wave interval parameter; or according to the harmonic interval parameter and the first harmonic offset parameter, the harmonic signal is obtained; according to the pitch parameter, the ratio between the harmonic signal and the noise signal is adjusted; and According to the adjusted harmonic signal and noise signal, the reconstructed spectral signal is obtained; that is, the spectral signal of the high frequency band is reconstructed by using an artificial reconstruction method based on the pitch parameter, the harmonic interval parameter and the first harmonic offset parameter; It is noted that when the first harmonic offset parameter is not transmitted in the coded code stream, the first harmonic offset parameter at the decoding end is equal to the harmonic interval parameter.

Perform frequency domain shaping on the reconstructed spectral signal according to the decoded frequency domain envelope, for example, perform frequency domain denormalization processing, and transform the shaped spectral signal into the time domain; it can be transformed by inverse MDCT or by inverse The FFT transformation transforms the trimmed spectral signal into the time domain, but it must correspond to the transformation method used by the encoding end;

Perform time-domain shaping processing, such as time-domain denormalization processing, according to the decoded time-domain envelope parameters, to obtain high-frequency signals decoded from parametric audio; and obtain synthesized audio signals.

It should be noted that the order of frequency-domain shaping and time-domain shaping is not unique, that is, time-domain shaping may be performed on the reconstructed spectrum signal first, and then frequency-domain shaping may be performed. For example: performing frequency-domain shaping processing on the reconstructed spectral signal according to the frequency-domain envelope parameters to obtain a frequency-domain shaped signal, and performing time-domain shaping on the frequency-domain shaped signal according to the time-domain envelope parameters processing to obtain a synthesized audio signal; or, perform time-domain shaping processing on the reconstructed spectrum signal according to the time-domain envelope parameters to obtain a time-domain shaped signal, and perform time-domain shaping according to the frequency-domain envelope parameters The final signal is subjected to frequency-domain shaping processing to obtain a synthetic audio signal.

The above content describes that when the audio signal is encoded by frequency division, it is judged whether the spectral signal of the audio signal in the current frequency band is similar to the spectral signal of the audio signal in the previous frequency band, and when they are not similar, the time domain envelope parameters are extracted , frequency domain envelope parameters, tone parameters, harmonic interval parameters and a set of parameters of the first harmonic offset parameters, when similar, only extract a set of time domain envelope parameters, frequency domain envelope parameters and tone parameters group parameters, or only extract a group of parameters including time-domain envelope parameters and frequency-domain envelope parameters, because the embodiment of the present invention reduces the number of parameters required for encoding, and at the same time reduces the number of parameters required for encoding. The number of bits required; it also effectively utilizes the similarity of the spectrum between different frequency bands of the signal to further reduce the coding rate and obtain a larger coding bandwidth. According to the above parameters, the decoding end can implement different spectral signal reconstruction methods according to the characteristics of different signals in the process of decoding audio signals in frequency bands, which is more adaptable to signal characteristics and can obtain the same high synthesis quality for different signals.

To facilitate the understanding of the embodiments of the present invention, the specific implementation solutions of encoding and decoding of the embodiments of the present invention will be described in detail below.

Embodiment two

In this embodiment, the input audio signal is divided into a high frequency band signal and a low frequency band signal at the encoding end, and encoding processing is performed on the high frequency band signal and the low frequency band signal respectively.

Fig. 5 is a schematic diagram of the processing process at the encoding end of Embodiment 2 of the present invention. As shown in Fig. 5, the encoding process includes:

51: Filter and analyze the input audio signal; set the sampling rate of the input audio signal to 32KHz, and the processing frame length is 20ms; after the input signal is divided into frequency bands and down-sampled, the signals corresponding to the 0~8kHz frequency band are 320 sampling points, corresponding to 8~16kHz frequency band signal has 320 sampling points;

52: Signals in the 0~8kHz frequency band are coded by core codes. In specific applications, core codes can be completed by G.729.1 codecs or by other broadband signal codecs, that is, no matter what codes are used In this way, it is enough to encode the signal in the 0~8kHz frequency band; and output the bit stream of the low frequency signal, that is, the output code stream;

53: For signals in the 8-16kHz frequency band, such as time-domain signals {y_hi(0), y_hi(1), ..., y_hi(319)}, use the encoding processing method provided by the embodiment of the present invention to perform parametric audio encoding: Here the high-frequency band is the current frequency band described in the encoding processing method, and the low-frequency band is the previous frequency band; when the spectrum of the high-frequency signal is not similar to the spectrum of the low-frequency signal, extracting the parameter, frequency domain envelope parameter, tone parameter, harmonic interval parameter, first harmonic offset parameter and a set of parameters of coding mode parameter; when there is similarity, only extract including time domain envelope parameter, frequency domain Envelope parameters, tone parameters and encoding mode parameters can also only extract a group of parameters including time domain envelope parameters, frequency domain envelope parameters and encoding mode parameters; the specific processing process can include:

(1) Determine the coding mode parameter CM; specifically, the cross-correlation between the low-band signal spectrum and the high-band signal spectrum can be calculated first to determine the similarity between the low-band harmonic structure and the high-band harmonic structure ; When the cross-correlation is greater than a certain threshold value, it can be judged that the harmonic structure of the low frequency band and the harmonic structure of the high frequency band are similar, set CM to 1, and use the spectral copy shaping method to pass the spectral signal of the low frequency band Reconstruct the spectral signal of the high frequency band; or reconstruct the spectral signal by other methods different from the spectral replication; when the cross-correlation is less than or equal to the threshold value, it is determined that the harmonic structure of the low frequency band and the harmonic structure of the high frequency band are inconsistent Similarly, set CM to 0, and artificially reconstruct the spectral signal in the high frequency band according to the parameters; of course, in practical applications, a simple method can also be used to determine the coding mode, that is, when the harmonic interval PG is less than a certain When a domain value, set CM to 1; otherwise set to 0;

(2) Calculate the subframe energy envelope {temp_env (0), temp (1), ..., temp (N-1)} and the global gain factor gain of the signal, N=8 in this embodiment; and according to this Two sets of values determine whether the signal is a steady-state signal or a transient signal; if it is a steady-state signal, the global gain factor gain is quantized, and the obtained quantized value is used as a time-domain envelope parameter, and encoded and written into the code stream; if If it is a transient signal, the energy envelope of the subframe is quantized, and the obtained quantized value is used as the time domain envelope parameter, and is encoded and written into the code stream; Domain normalization processing to obtain the time domain normalized signal;

(3) After the time-domain normalized signal is transformed by MDCT (modified discrete cosine transform, Modified Discrete CosineTransform) (for example, 640 points), a set of MDCT coefficients is obtained, that is, the frequency domain signal corresponding to the frequency band {y_swb(0 ), y_swb(1),..., y_swb(319)}, since the UWB coder only requires to process signals in the 8~14kHz frequency band, it only processes {y_swb(0), y_swb(1), ..., y_swb(239)} part; during processing, this group of frequency domain signals is divided into N subbands, and the offspring energy of each subband is extracted and quantized to obtain a set of quantized frequency domain envelope {spec_env(0) , spec_env(1),..., spec_env(N-1)}, which is the frequency domain envelope parameter in the 8~14kHz frequency band;

As for the broadband core encoder G.729.1, the 7~8kHz part of the signal is no longer within its processing range, in order to ensure the continuity of the decoded signal spectrum at the decoder, it is also necessary to extract the characteristic parameters of the 7~8kHz part of the signal; due to G. The 729.1 encoder performs MDCT transformation (for example, 320 points) on the 4~8kHz signal, and the corresponding frequency domain signal {y_wb(0), y_wb(1),...,y_wb(159)}, where 7~8kHz corresponds to The frequency domain signal is {y_wb(120), y_wb(121),...,y_wb(159)}, which is divided into M subbands, and the frequency domain envelope of each subband is extracted and quantized to obtain a set of 7~8kHz The quantized frequency domain envelope in the frequency band {spec_env_extra(0), spec_env_extra(1), ..., spec_env_extra(M-1)}, together with the frequency domain envelope parameters in the 8~14kHz frequency band, form the entire frequency domain Envelope parameters; this group of envelopes can be transmitted to the decoding end after encoding; in this embodiment, N=15, M=3;

(4) Extract tone parameters; specifically, parameter extraction can be performed directly in the MDCT domain; in order to further improve the performance of the encoder, it is also possible not to directly perform parameter extraction in the MDCT domain, but according to the original frequency domain signal {y_swb(0) , y_swb(1), ..., y_swb(239)} calculate the pseudo-spectral signal, and calculate the tone parameter according to the pseudo-spectral signal;

Concrete pseudo spectrum signal S(k)={S(0), S(1),..., S(239)} can be calculated according to the following formula:

$\mathrm{<span\; class="notranslate">\; S</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\left\{\begin{array}{c}\sqrt{\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; \_</span>{\mathrm{<span\; class="notranslate">\; swb</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; \_</span>{\mathrm{<span\; class="notranslate">\; swb</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}\\ \sqrt{\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; \_</span>{\mathrm{<span\; class="notranslate">\; swb</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 239</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; \_</span>{\mathrm{<span\; class="notranslate">\; swb</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 238</span>}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 239</span>}\\ \sqrt{\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; \_</span>{\mathrm{<span\; class="notranslate">\; swb</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; \_</span>\mathrm{<span\; class="notranslate">\; swb</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; \_</span>\mathrm{<span\; class="notranslate">\; swb</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; otherwise</span>}\end{array}\right.,<span\; class="notranslate">\; ,</span>$

Of course, other methods can also be used, such as {|y_swb(0)|, |y_swb(1)|, ..., |y_swb(239)|} obtained by directly taking the absolute value of the original frequency domain signal; then calculate from Correlation function ACF(k ₀ ), the autocorrelation function can be calculated from the pseudo-spectral signal in the frequency domain, for example, ACF(k ₀ )=IFFT(|FFT(S(k))| ² ), where FFT is Fast Fourier Transform, IFFT is its inverse transformation; in addition, it can also be directly calculated, for example $\mathrm{ACF}\left(k_0\right)=\underset{k=0}{\overset{239}{\mathrm{\&Sigma;}}}S\left(k\right)S(k+k_0);$ In addition, the average amplitude difference function (AMDF) can also be used to enhance the autocorrelation function;

The tone parameter can be represented by the ratio between the maximum value and the minimum value of the autocorrelation function, such as T=max(ACF(k ₀ ))/min(ACF(k ₀ )), where the maximum value and the minimum value are obtained at within the expected range or within the beneficial range for the calculation of harmonic spacing parameters;

(5) According to ACF(k ₀ ), estimate the harmonic interval parameter PG; the harmonic interval parameter of the high-frequency band signal is usually extracted in the frequency domain (or transform domain); the integer value of the harmonic interval can be obtained by the peak The extraction method is estimated by the autocorrelation function, for example, according to PG=argmax(ACF(k ₀ )), where the maximum value can be obtained within a desired range or a range of interest, and the harmonic interval The fractional value of can be obtained by peak extraction method after interpolating the autocorrelation function ACF(k ₀ ) appropriately; it is also possible to interpolate the autocorrelation function only around the obtained integer harmonic interval, and then extract the peak value The method to obtain the fractional value of the harmonic interval;

(6) The estimated harmonic interval parameter value can also be corrected to suppress the generation of multiplier and fractional frequency; for example, compare the obtained harmonic interval PG of the current frame with the harmonic interval old_PG of the previous frame , if the ratio between the harmonic interval of the current frame and the harmonic interval of the previous frame is less than a certain threshold (such as 0.1) and ACF(old_PG)>0.95ACF(PG), replace it with the harmonic interval of the previous frame Harmonic interval PG obtained in this frame=old_PG;

其中

表示向下取整；需要指出的是，实际上谐波间隔参数与第一谐波偏移量参数之间也存在着一定程度上的相关性，因此可以通过谐波间隔参数估计出高频带信号的第一谐波偏移量参数；同时，第一谐波偏移量参数也可以用来进一步修正谐波间隔参数的估计值，从而达到更优的参数提取效果；(7) Estimate the first harmonic offset parameter P0 according to the harmonic interval; for example, specifically, within the range of the harmonic interval, that is, within the range of [0, PG], the first harmonic component can be placed in different Offset the position, and place other harmonics in sequence according to the harmonic interval, and calculate the correlation between the resulting spectrum and the pseudo-spectrum, the offset position with the greatest correlation is the first harmonic offset sought, For example

in

Indicates rounding down; it should be pointed out that there is actually a certain degree of correlation between the harmonic interval parameter and the first harmonic offset parameter, so the high frequency band can be estimated by the harmonic interval parameter The first harmonic offset parameter of the signal; at the same time, the first harmonic offset parameter can also be used to further modify the estimated value of the harmonic interval parameter, so as to achieve a better parameter extraction effect;

(8) When CM is equal to 1, a group of parameters including encoding mode parameters, time domain envelope parameters, frequency domain envelope parameters, and tone parameters will be quantized or encoded, and transmitted to the decoding end (that is, the transmission of high frequency Parameter bit stream); when CM is equal to 0, a group of parameters including encoding mode parameters, time domain envelope parameters, frequency domain envelope parameters, tone parameters, harmonic interval parameters and first harmonic offset parameters, will be quantized or encoded, and transmitted to the decoding end (that is, transmit high-frequency parameter bit stream);

It should be pointed out that when CM is equal to 1, only a set of parameters including encoding mode parameters, time domain envelope parameters and frequency domain envelope parameters may be quantized or encoded, and transmitted to the decoding end;

54: After completing the parametric audio coding of the high-frequency band signal, it is possible to choose whether to use the optional RIRAC audio coding to enhance the high-frequency signal after parametric audio coding according to the number of remaining coding bits; the enhancement method adopted in this embodiment It is to transform and encode the high-frequency band signal in the MDCT domain; of course, other methods can also be used to enhance the high-frequency signal after parametric audio coding, such as the high-band original signal and the residual signal after high-band audio coding. Transform coding, etc.; and transmit high-frequency enhanced bit streams.

Correspondingly, after receiving the above-mentioned low-frequency bit stream, high-frequency parameter bit stream, and high-frequency enhanced bit stream, the decoder performs decoding and synthesizes an audio signal; FIG. As shown in 6, the specific decoding process may include:

61: The signal synthesis in the 0~8kHz frequency band is completed through core decoding;

62: The signal synthesis in the 8~16kHz frequency band is completed through parameter audio decoding; the specific processing includes: (1) decoding the encoding mode parameter CM according to the received data;

(2) Decoding time domain envelope parameters and frequency domain envelope parameters from the data;

(3) If CM is equal to 1, the tone parameters can be decoded from the received data, and the spectrum signal of the high frequency band can be reconstructed through the spectrum of the low frequency band by means of spectrum copy shaping, or reconstructed by other methods different from spectrum copy Spectral signal; for example, specifically, according to the tone parameter, the spectral signal of the low-band signal obtained through core decoding can be shaped, and the shaped spectral signal can be used as the reconstructed high-band spectral signal;

It should be pointed out that when the encoding end does not transmit tone parameters when CM is equal to 1, the decoding end will directly use the low-band spectral signal obtained by core decoding as the reconstructed high-band spectral signal;

If CM is equal to 0, the tone parameter, harmonic interval parameter and first harmonic offset parameter can be decoded from the received data, and the parameters based on the tone parameter, harmonic interval parameter and first harmonic offset can be used The manual reconstruction method to reconstruct the spectral signal of the high frequency band; the reconstruction method of the spectral signal is based on the harmonic signal plus the noise signal; specifically, the harmonic with random phase is placed in the form of pulses at certain frequencies in the frequency domain Point above, thereby reconstructing the harmonic signal, where the pulse interval is determined by the harmonic interval parameter, the position of the first pulse can be obtained according to the first harmonic offset; the noise signal can be obtained by a random number generator; according to The value of the tone parameter T adjusts the ratio between the harmonic signal and the noise signal; and adds the adjusted harmonic signal and the noise signal to obtain the reconstructed spectral signal; there are various specific adjustments, for example: first separate Calculate the energy of the harmonic signal and the noise signal, denoted as enerP and enerN, and then calculate the adjustment factor β ₁ =1-T and ${\mathrm{\&beta;}}_2=\sqrt{\frac{\mathrm{enerP}}{\mathrm{ener\; N}}}*T,$ and get the reconstructed spectral signal $\hat{S}\left(k\right)={\mathrm{\&beta;}}_1\mathrm{buf}\_\mathrm{pulses}\left(k\right)+{\mathrm{\&beta;}}_2\mathrm{buf}\_\mathrm{noise}\left(k\right);$

(4) Perform frequency-domain shaping on the reconstructed spectral signal according to the decoded frequency-domain envelope, such as frequency-domain denormalization processing, and transform the shaped spectral signal into the time domain; for example, it can be changed by inverse MDCT, The trimmed spectral signal can also be transformed to the time domain by inverse FFT transformation;

(5) Perform time-domain shaping processing according to the decoded time-domain envelope parameters, such as time-domain denormalization processing, to obtain decoded high-frequency signals;

It should be noted that, in the denormalization process of the time domain and the frequency domain, an optional smoothing filter process can also be performed on the time domain envelope and the frequency domain envelope. If the spectral signal in the high frequency band is artificially reconstructed, once the harmonics are placed in the wrong subband, the wrong envelope factor will be used for denormalization. If there is a slight deviation in the harmonic position, a certain degree of distortion will be introduced, which can be mitigated by smoothing filtering. Specifically, if there is a very strong tonal component near the subband boundary, then the interpolated subband energy envelope factor can be used for frequency domain denormalization; then the obtained signal is transformed into the time domain, and then interpolate the time domain gain function in the time domain by the adaptive subframe energy envelope (ATE); this time domain gain function can finally be used to denormalize the time domain signal;

63: After completing the decoding of the high-frequency signal at 62, you can choose whether to enhance the encoded high-frequency signal according to the number of bits left in the received data. The specific method corresponds to the enhancement method adopted by the encoding end. I won't go into details here;

64: Combine the synthesized signal in the 0~8kHz frequency band and the synthesized signal in the 8~16kHz frequency band through QMF synthesis filtering to obtain the final synthesized audio signal with a sampling rate of 32kHz.

In the second embodiment, when the audio signal is divided into a low frequency band signal and a high frequency band signal, parametric encoding and decoding are performed on the high frequency band signal, that is, the encoding mode parameter is used to indicate the use of the included time domain of the representative signal Envelope, frequency domain envelope, tone, harmonic interval and first harmonic offset to complete the encoding and decoding, or use a set of time domain envelope, frequency domain envelope and tone to characterize the signal parameters to complete the encoding and decoding. A set of parameters adopted in the embodiment of the present invention reduces the number of parameters required for encoding, and simultaneously reduces the number of bits required for encoding using parameters; thereby solving the problem of high bit numbers in existing encoding methods; at the same time , compared with the existing parametric audio coding algorithm, because this group of parameters in the embodiment of the present invention can be coded with fewer bits, thereby further reducing the coding rate of the signal, and when the transmission capacity of the channel is constant, because the wooden The invention has a low number of coding bits, so it can code a signal with a higher bandwidth, and achieve a larger coding bandwidth and higher coding quality with a lower coding rate. At the same time, the audio signal can be synthesized with fewer bits at the decoding end, and the quality of the audio signal is high; and, when the harmonic structure of the audio signal is obvious, the audio quality obtained through decoding is better.

Embodiment Three

Compared with Embodiment 2, which adopts the method of firstly extracting the time-domain envelope parameters and then extracting the frequency-domain envelope parameters, Embodiment 3 adopts the method of first extracting the frequency-domain envelope parameters to realize encoding (based on the audio frequency domain in Embodiment 3). The signal and frequency division method are the same as those in Embodiment 2 as an example).

In this embodiment, the process of processing the high frequency band signal at the encoding end may specifically include:

(1): Determine the encoding mode parameter CM according to the method in (1) of the encoding end in Embodiment 2;

(2): The time domain signal in the 8~16kHz frequency band is transformed by MDCT to obtain a set of MDCT coefficients. Since the ultra-wideband part only processes the signal in the 8~14kHz frequency band, only {y_swb(0) is processed for the frequency domain signal , y_swb(1),...,y_swb(239)} part; for the core code, the 7~8kHz part of the signal is no longer within its processing range, in order to ensure the continuity of the decoded signal spectrum at the decoder, it is necessary to extract 7~8kHz partial MDCT transform domain signal {y_wb(120), y_wb(121),...,y_wb(159)};

(3): The MDCT coefficients in the 7~14kHz frequency band are divided into bands, and the respective sub-band energies are calculated as frequency domain envelope parameters, and quantized and encoded for transmission;

(4): Perform frequency domain normalization processing on the MDCT coefficients in the 7~14kHz frequency band, and extract linear prediction coefficients according to the MDCT coefficients after frequency domain normalization, as time domain envelope parameters, and perform linear prediction on this group The coefficients are quantized and then encoded and transmitted;

(5): Perform linear prediction filtering on the MDCT coefficients normalized in the frequency domain to obtain the linear prediction residual in the MDCT domain;

(6): According to the method in (4)～(8) of encoding terminal 53 in embodiment two, extract the pitch parameter, harmonic interval parameter and first harmonic offset parameter of high-frequency signal; When encoding mode is When 1, only the encoding mode parameter, time domain envelope parameter, frequency domain envelope parameter and tone parameter are transmitted to the decoder; when the encoding mode is 0, the encoding mode parameter, time domain envelope parameter, frequency domain envelope parameters, tone parameters, harmonic interval parameters and first harmonic offset parameters are transmitted to the decoding end together;

Correspondingly, the process of processing the high frequency band signal at the decoding end may specifically include:

(7): According to the received code stream, decode the coding mode parameter CM; and decode the time-domain envelope parameters and frequency-domain envelope parameters from the code stream; specifically, the quantized linearity can be obtained by searching the codebook The prediction coefficient is an envelope parameter in the instant domain; in order to perform time-domain shaping processing according to the obtained linear prediction coefficient; the quantized sub-band energy is obtained through codebook search, that is, the envelope parameter in the frequency domain; frequency-domain shaping processing of the sub-band energy;

(8): reconstruct the spectral signal of the high frequency band according to the method in (3) in the decoding end 62 in the second embodiment;

(9): Make the reconstructed high-band spectrum signal pass through the linear prediction inverse filter, which is equivalent to performing time-domain shaping processing on the reconstructed high-band spectrum signal;

(10): according to the quantized sub-band energy, perform frequency-domain shaping processing on the reconstructed high-frequency band spectrum signal;

(11): Through the inverse MDCT transformation, transform the shaped high frequency band spectrum signal into the time domain to obtain the final high frequency band composite signal.

It can be seen from the above description that Embodiment 3 adopts the method of extracting the frequency-domain envelope parameters first to implement encoding. Since the order of obtaining the above-mentioned parameters is not unique, that is, no matter in what order, as long as the above-mentioned audio signal encoding mode parameters, Time domain envelope parameters, frequency domain envelope parameters, pitch parameters, harmonic interval parameters and first harmonic offset parameters are enough. A set of parameters adopted in the embodiment of the present invention reduces the number of parameters required for encoding, and simultaneously reduces the number of bits required for encoding using parameters; thereby solving the problem of high bit numbers in existing encoding methods; at the same time , compared with the existing parametric audio coding algorithm, since this group of parameters in the embodiment of the present invention can be coded with fewer bits, thereby further reducing the coding rate of the signal, and when the transmission capacity of the channel is constant, due to the The invention has a low number of coding bits, so it can code a signal with a higher bandwidth, and achieve a larger coding bandwidth and higher coding quality with a lower coding rate. At the same time, the audio signal can be synthesized with fewer bits at the decoding end, and the quality of the audio signal is high; and, when the harmonic structure of the audio signal is obvious, the audio quality obtained through decoding is better.

The embodiment of the present invention also provides a corresponding audio encoding device, the structure of which is shown in Figure 7, and the specific implementation structure may include:

A parameter extraction unit 71, configured to extract time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters and harmonic interval parameters used to characterize the audio signal; when the harmonic interval of the audio signal is offset from the first harmonic When the value of the quantity is different, it is also used to extract the first harmonic offset parameter used to characterize the audio signal, and transmit it to the sending unit;

The sending unit 72 is configured to encode the time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters and harmonic interval parameters, and transmit them to the decoding end. Specifically, for example: the time-domain envelope parameters, The frequency domain envelope parameters, tone parameters and harmonic interval parameters are encoded and then transmitted to the decoder; or used to transmit the time domain envelope parameters, frequency domain envelope parameters, tone parameters, harmonic interval parameters and the first A harmonic offset parameter is encoded and transmitted to the decoding end.

The embodiment of the present invention also provides a corresponding audio decoding device, the structure of which is shown in Figure 8, and the specific implementation structure may include:

The decoding unit 81 is used to decode the received data to obtain time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters and harmonic interval parameters used to characterize the audio signal; Decoding the data of a harmonic offset parameter to obtain a first harmonic offset parameter used to characterize the audio signal;

Synthesizing unit 82, is used for according to time domain envelope parameter, frequency domain envelope parameter, tone parameter and harmonic interval parameter; Or time domain envelope parameter, frequency domain envelope parameter, tone parameter, harmonic interval parameter and first Harmonic offset parameter, synthesized audio signal; specific can include:

The harmonic reconstruction subunit 821 is configured to obtain a harmonic signal according to the harmonic interval parameter; or when the harmonic interval used to characterize the audio signal is different from the first harmonic offset, according to the harmonic wave interval parameter and the first harmonic offset parameter to obtain a harmonic signal;

The spectral signal reconstruction subunit 822 is configured to adjust the ratio between the harmonic signal and the noise signal obtained by the harmonic reconstruction subunit 821 according to the tone parameter; and obtain reconstructed spectral signal;

The shaping subunit 823 is configured to process the spectral signal reconstructed by the spectral signal reconstruction subunit 822 according to the frequency domain envelope parameter and the time domain envelope parameter to obtain a synthesized audio signal; for example: according to the frequency domain envelope Carry out frequency-domain shaping processing on the reconstructed spectrum signal by the envelope parameters to obtain a signal after frequency-domain shaping, and perform time-domain shaping processing on the signal after frequency-domain shaping according to the time-domain envelope parameters to obtain the synthesized audio signal or, performing time-domain shaping processing on the reconstructed spectrum signal according to the time-domain envelope parameters to obtain a signal after time-domain shaping, and performing frequency-domain processing on the signal after time-domain shaping according to the frequency-domain envelope parameters shaping process to obtain the synthesized audio signal.

The embodiment of the present invention also provides a corresponding audio codec system, the structure of which is shown in Figure 9, and the specific implementation structure may include:

The encoding device 91 is used to extract time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters and harmonic interval parameters used to characterize audio signals; Envelope parameters, tone parameters and harmonic interval parameters are encoded and then sent to the decoding device; the details may include:

The parameter extraction unit 911 is used to extract the time domain envelope parameter, frequency domain envelope parameter, tone parameter and harmonic interval parameter of the audio signal; when the harmonic interval of the audio signal and the value of the first harmonic offset At the same time, it is also used to extract the first harmonic offset parameter of the audio signal;

The sending unit 912 is configured to send the time-domain envelope parameter, frequency-domain envelope parameter, tone parameter, and harmonic interval parameter; or the time-domain envelope parameter, frequency-domain envelope parameter, tone parameter, and harmonic interval The parameter and the first harmonic offset parameter are encoded and transmitted to the decoding device;

The decoding device 92 is configured to decode the data sent by the encoding device to obtain the time domain envelope parameter, frequency domain envelope parameter, tone parameter and harmonic interval parameter; according to the time domain envelope parameter, Frequency domain envelope parameters, tone parameters and harmonic interval parameters synthesize audio signals; specifically, it may include:

The decoding unit 921 is configured to decode the received data to obtain the time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters and harmonic interval parameters, or the time-domain envelope parameters and frequency-domain envelope parameters parameter, pitch parameter, harmonic spacing parameter and first harmonic offset parameter;

Synthesizing unit 922, configured to use the time domain envelope parameter, frequency domain envelope parameter, tone parameter and harmonic interval parameter, or the time domain envelope parameter, frequency domain envelope parameter, tone parameter, harmonic interval parameter and first harmonic offset parameter, synthesized audio signal.

The embodiment of the present invention also provides a corresponding encoding processing device, the structure of which is shown in Figure 10, and the specific implementation structure may include:

Judging unit 101, configured to judge whether the spectral signal of the audio signal in the current frequency band is similar to the spectral signal of the audio signal in the previous frequency band; specifically, the value of the encoding mode parameter can be used to indicate whether the information is similar;

The encoding unit 102 is configured to extract the time-domain packets used to characterize the audio signal when the spectral signal of the audio signal in the current frequency band is similar to the spectral signal of the audio signal in the previous frequency band according to the judgment result information obtained by the judgment unit 101 Envelope parameters and frequency domain envelope parameters are also used to extract tone parameters; or, when the spectral signal of the audio signal in the current frequency band is not similar to the spectral signal between the audio signal in the previous frequency band, extract the time used to characterize the audio signal Domain envelope parameters, frequency domain envelope parameters, pitch parameters and harmonic interval parameters; when the harmonic interval of the audio signal is different from the value of the first harmonic offset, it is also used to extract the audio signal First Harmonic Offset parameter;

The transmission unit 103 is used to send information similar to the spectral signal between the spectral signal of the audio signal of the current frequency band obtained by the judging unit 101 and the spectral signal of the audio signal of the previous frequency band, such as encoding the encoding mode parameter and sending it; The time-domain envelope parameters and frequency-domain envelope parameters (may also include tone parameters) of the audio signal extracted by the encoding unit are encoded and sent; or, the spectrum of the audio signal in the current frequency band obtained by the judging unit is sent Information about the spectral signal dissimilarity between the signal and the audio signal of the previous frequency band, the time domain envelope parameter, frequency domain envelope parameter, pitch parameter and harmonic interval parameter of the audio signal extracted by the encoding unit (may also include The first harmonic offset parameter) is encoded and sent.

The embodiment of the present invention also provides a corresponding decoding processing device, the structure of which is shown in Figure 11, and the specific implementation structure may include:

The receiving information unit 111 is used to receive information indicating that the spectral signal of the audio signal in the current frequency band is similar to the spectral signal of the audio signal in the previous frequency band, and decode the received data to obtain time-domain envelope parameters used to characterize the audio signal and frequency domain envelope parameters; or, receiving information indicating that the spectral signal of the audio signal of the current frequency band is not similar to the spectral signal of the audio signal of the previous frequency band, and decoding the received data to obtain the time domain used to characterize the audio signal Envelope parameters, frequency domain envelope parameters, pitch parameters and harmonic interval parameters; also used to decode the data containing the first harmonic offset parameter to obtain the first harmonic offset parameter used to characterize the audio signal Specifically, the receiving information unit 111 may determine whether the spectral signal of the audio signal of the current frequency band is similar or dissimilar to the spectral signal of the audio signal of the previous frequency band according to the received encoding mode parameter;

The decoding unit 112 is configured to synthesize an audio signal according to the similar information received by the receiving information unit 111, and the time-domain envelope parameters and frequency-domain envelope parameters used to characterize the audio signal; or, according to the The above dissimilar information, and the time domain envelope parameters, frequency domain envelope parameters, pitch parameters and harmonic interval parameters used to characterize the audio signal, synthesize the audio signal; specifically:

When receiving information that the spectral signal of the audio signal in the current frequency band is similar to the spectral signal of the audio signal in the previous frequency band, the decoding unit 112, specifically as shown in FIG. 12 , may include:

Reconstruction subunit 121: used to reconstruct the spectrum signal to obtain the reconstructed spectrum signal;

The second shaping subunit 122: for performing shaping processing on the spectral signal reconstructed by the reconstruction subunit 121 according to the pitch parameter, to obtain the reconstructed spectral signal after shaping;

The first shaping subunit 123: used to process the reconstructed spectral signal (or shaped spectral signal) according to the frequency domain envelope parameter and time domain envelope parameter to obtain a synthesized audio signal; for example: according to the The frequency-domain envelope parameter and the time-domain envelope parameter process the reconstructed spectrum signal after the shaping process by the second shaping subunit, including: performing frequency-domain processing on the reconstructed spectrum signal according to the frequency-domain envelope parameter. Shaping processing to obtain a frequency-domain shaped signal; performing time-domain shaping processing on the frequency-domain shaped signal according to the time-domain envelope parameter to obtain the synthesized audio signal; or, according to the time-domain envelope parameter. performing time-domain shaping processing on the reconstructed spectrum signal to obtain a time-domain shaping signal; performing frequency-domain shaping processing on the time-domain shaping signal according to the frequency domain envelope parameters to obtain the synthesized audio signal;

When receiving information that the spectral signal of the audio signal in the current frequency band is not similar to the spectral signal of the audio signal in the previous frequency band, the decoding unit 112, specifically as shown in FIG. 12 , may include:

The harmonic reconstruction subunit 124 is configured to obtain a harmonic signal according to the harmonic interval parameter; or obtain a harmonic signal according to the harmonic interval parameter and the first harmonic offset parameter;

The spectral signal reconstruction subunit 125 is configured to adjust the ratio between the harmonic signal and the noise signal according to the tone parameter, and obtain a reconstructed spectral signal according to the adjusted harmonic signal and noise signal;

The third shaping subunit 126 is configured to process the reconstructed spectrum signal according to the frequency domain envelope parameter and the time domain envelope parameter to obtain a synthesized audio signal. For example, performing frequency domain shaping processing on the reconstructed spectral signal according to the frequency domain envelope parameters to obtain a frequency domain shaped signal; performing time domain shaping on the frequency domain shaped signal according to the time domain envelope parameters Processing to obtain the synthesized audio signal; or, performing time-domain shaping processing on the reconstructed spectrum signal according to the time-domain envelope parameters to obtain a signal after time-domain shaping; The signal after domain shaping is subjected to frequency domain shaping processing to obtain a synthetic audio signal.

The various embodiments of the present invention above can be applied to audio codec devices, but are not limited to.

To sum up, compared with the prior art, each embodiment of the present invention adopts the parameters including time-domain envelope parameter, frequency-domain envelope parameter, pitch parameter and harmonic interval parameter (may also include the first Harmonic Offset Parameter) is a set of parameters to represent the audio signal. When encoding the audio signal, it can reduce the number of bits required for encoding using parameters on the existing basis, and can use fewer bits. Encode the signal to further reduce the encoding rate of the signal, thereby obtaining a larger encoding bandwidth and higher encoding quality with a lower encoding rate, especially for signals with obvious harmonic structures, the embodiment of the present invention can obtain a lot of Good encoding quality. At the same time, in the encoding and decoding processing technical solutions provided by the embodiments of the present invention, when the audio signal is encoded in a frequency-divided manner, it is judged whether the spectral signal of the audio signal of the current frequency band is similar to the spectral signal of the audio signal of the previous frequency band, When dissimilar, extract a set of parameters including time domain envelope parameters, frequency domain envelope parameters, pitch parameters and harmonic interval parameters (may also include the first harmonic offset parameter), and when similar, only extract the parameters containing A set of parameters including domain envelope parameters and frequency domain envelope parameters (which may also include tone parameters) effectively utilizes the similarity of spectra between different frequency bands of the signal to further reduce the coding rate and obtain a larger coding bandwidth. According to the above parameters, the decoding end can implement different spectral signal reconstruction methods according to the characteristics of different signals in the process of decoding audio signals in frequency bands, which is more adaptable to signal characteristics and can obtain the same high synthesis quality for different signals. In other words, when the transmission capability of the channel is constant, the present invention can encode a signal with a higher bandwidth because the number of encoding bits is lower. In terms of auditory perception, the larger the bandwidth of the signal, the better the auditory experience. Therefore, when the transmission capacity of the channel is constant, the method provided by the present invention can obtain higher coding bandwidth and higher synthesis quality. In addition, the embodiment of the present invention provides a technical solution for sub-band encoding and decoding processing of audio signals, which can achieve a larger encoding bandwidth with a lower encoding rate and obtain a higher encoding bandwidth during the process of encoding and decoding audio signals in sub-bands. High encoding quality.

The above is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed in the present invention can easily think of changes or Replacement should be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be determined by the protection scope of the claims.

Claims (27)

1. An audio coding method, characterized in that, comprising: Extract time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters and harmonic interval parameters used to characterize the audio signal, and the pitch parameters represent the ratio between the harmonic signal and the noise signal in the audio signal; After encoding the time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters and harmonic interval parameters, transmit them to the decoding end; The method further includes: when the harmonic interval of the audio signal is different from the value of the first harmonic offset, further extracting the first harmonic offset parameter used to characterize the audio signal, and comparing After encoding, it is transmitted to the decoding end, and the first harmonic offset parameter represents the position of the first harmonic of the audio signal. 2. An audio coding device, characterized in that, comprising: A parameter extraction unit is used to extract time-domain envelope parameters, frequency-domain envelope parameters, tone parameters and harmonic interval parameters used to characterize the audio signal, and the tone parameters represent the ratio between the harmonic signal and the noise signal in the audio signal; A sending unit, configured to encode the time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters and harmonic interval parameters, and transmit them to the decoding end; The parameter extraction unit is also used for: When the harmonic interval of the audio signal is different from the value of the first harmonic offset, the first harmonic offset parameter used to characterize the audio signal is further extracted and sent to the sending unit, the first A harmonic offset parameter characterizes the position of the first harmonic of the audio signal; The sending unit is further configured to encode the first harmonic offset parameter and transmit it to the decoding end. 3. An audio decoding method, characterized in that, comprising: Decode the received data to obtain the time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters and harmonic interval parameters used to characterize the audio signal. The pitch parameter represents the gap between the harmonic signal and the noise signal in the audio signal. Proportion; According to the time-domain envelope parameter, the frequency-domain envelope parameter, the pitch parameter and the harmonic interval parameter, synthesize an audio signal, and the step of synthesizing the audio signal includes: obtaining a harmonic signal according to the harmonic interval parameter; adjusting the ratio between the harmonic signal and the noise signal according to the tone parameter; obtaining a reconstructed spectrum signal according to the adjusted harmonic signal and noise signal; processing the reconstructed spectrum signal according to the frequency domain envelope parameter and the time domain envelope parameter to obtain a synthesized audio signal; The method further includes: decoding the received data containing the first harmonic offset parameter to obtain the first harmonic offset parameter used to characterize the audio signal, the first harmonic offset parameter The position of the first harmonic of the audio signal is characterized, and the steps of synthesizing the audio signal include: obtaining a harmonic signal according to the harmonic interval parameter and the first harmonic offset parameter; adjusting the ratio between the harmonic signal and the noise signal according to the tone parameter; obtaining a reconstructed spectrum signal according to the adjusted harmonic signal and noise signal; Processing the reconstructed spectrum signal according to the frequency domain envelope parameter and the time domain envelope parameter to obtain a synthesized audio signal. 4. The method according to claim 3, further comprising: when the harmonic interval used to characterize the audio signal is different from the first harmonic offset parameter, receiving The data of the wave offset parameter is decoded to obtain the first harmonic offset parameter used to characterize the audio signal. 5. The method according to claim 3 or 4, wherein said processing the reconstructed spectrum signal according to said frequency domain envelope parameter and time domain envelope parameter to obtain a synthetic audio signal comprises: performing frequency domain shaping processing on the reconstructed spectral signal according to the frequency domain envelope parameters to obtain a frequency domain shaped signal, and performing time domain shaping processing on the frequency domain shaped signal according to the time domain envelope parameters, obtaining the synthesized audio signal; or, performing time domain shaping processing on the reconstructed spectrum signal according to the time domain envelope parameters to obtain a time domain shaped signal, and performing frequency domain shaping processing on the time domain shaped signal according to the frequency domain envelope parameters, The synthesized audio signal is obtained. 6. An audio decoding device, characterized in that, comprising: The decoding unit is used to decode the received data to obtain time-domain envelope parameters, frequency-domain envelope parameters, tone parameters and harmonic interval parameters used to represent the audio signal, and the tone parameters represent the harmonic signal and harmonic interval parameters in the audio signal. The ratio between the noise signal; A synthesis unit, for synthesizing an audio signal according to the time-domain envelope parameter, the frequency-domain envelope parameter, the pitch parameter and the harmonic interval parameter; the synthesis unit includes: A harmonic reconstruction subunit, configured to obtain a harmonic signal according to the harmonic interval parameter; or when the harmonic interval used to characterize the audio signal is different from the first harmonic offset, according to the harmonic interval parameter and the first harmonic offset parameter to obtain a harmonic signal; The spectral signal reconstruction subunit is used to adjust the ratio between the harmonic signal and the noise signal obtained by the harmonic reconstruction subunit according to the tone parameter; and obtain the reconstructed spectral signal; A shaping subunit, configured to process the reconstructed spectrum signal according to the frequency-domain envelope parameter and the time-domain envelope parameter, to obtain a synthesized audio signal; The decoding unit is also used for: Decoding the received data containing the first harmonic offset parameter to obtain the first harmonic offset parameter used to characterize the audio signal, the first harmonic offset parameter characterizes the first harmonic offset parameter of the audio signal position of the harmonic. 7. An audio codec system, characterized in that, comprising: The encoding device according to claim 2; And the decoding device as claimed in claim 6. 8. An encoding processing method, characterized in that, comprising: When the audio signal is encoded in the way of frequency division, if the spectral signal of the audio signal in the current frequency band is similar to the spectral signal of the audio signal in the previous frequency band, the time domain envelope parameters and frequency domain parameters used to characterize the audio signal are extracted. Envelope parameters, encoding the time-domain envelope parameters and frequency-domain envelope parameters, and sending information indicating that the spectral signal of the audio signal of the current frequency band is similar to the spectral signal of the audio signal of the previous frequency band; If the spectral signal of the audio signal of the current frequency band is not similar to the spectral signal of the audio signal of the previous frequency band, the time domain envelope parameter, frequency domain envelope parameter, pitch parameter and harmonic interval parameter used to characterize the audio signal are extracted, Encoding the time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters and harmonic interval parameters and sending them, and simultaneously sending a signal indicating that the spectrum signal of the audio signal of the current frequency band is similar to the spectrum signal of the audio signal of the previous frequency band Information, the tone parameter characterizes the ratio between the harmonic signal and the noise signal in the audio signal; The information representing the similarity or dissimilarity between the spectral signal of the audio signal in the current frequency band and the spectral signal of the audio signal in the previous frequency band is specifically represented by a coding mode parameter. 9. The method according to claim 8, characterized in that, the method further comprises: The encoding mode parameter is used to indicate that when the spectral signal of the audio signal in the current frequency band is similar to the spectral signal of the audio signal in the previous frequency band, the decoding end will use the time domain envelope parameter and the frequency domain envelope parameter of the audio signal When decoding the audio signal of the current frequency band, and indicating that the spectral signal of the audio signal of the current frequency band at the decoding end is not similar to the spectral signal of the audio signal of the previous frequency band, according to the time domain envelope parameter used to characterize the audio signal , frequency domain envelope parameter, pitch parameter and harmonic interval parameter, to decode the audio signal of the current frequency band. 10. The method according to claim 8, characterized in that, the method further comprises: when extracting time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters and harmonic interval parameters used to characterize audio signals , when the harmonic interval of the audio signal is different from the value of the first harmonic offset, further extracting the first harmonic offset parameter of the audio signal, encoding it and sending it. 11. The method according to claim 8, characterized in that, the method further comprises: when extracting the time-domain envelope parameters and frequency-domain envelope parameters used to characterize the audio signal, also extracting the time-domain envelope parameters of the audio signal Tone parameters, encoded and sent. 12. An encoding processing device, comprising: A judging unit, configured to judge whether the spectral signal representing the audio signal of the current frequency band is similar to the spectral signal of the audio signal of the previous frequency band; The coding unit is used to extract the time-domain envelope parameters used to characterize the audio signal when the spectral signal of the audio signal in the current frequency band is similar to the spectral signal of the audio signal in the previous frequency band according to the judgment result information obtained by the judgment unit and frequency domain envelope parameters; or, when the spectral signal of the audio signal of the current frequency band is not similar to the spectral signal between the audio signal of the previous frequency band, the time domain envelope parameters and the frequency domain envelope used to characterize the audio signal are extracted Parameters, tone parameters and harmonic interval parameters, the tone parameters characterize the ratio between the harmonic signal and the noise signal in the audio signal; the spectral signal representing the audio signal of the current frequency band is similar to the spectral signal of the audio signal of the previous frequency band or Dissimilar information, specifically expressed by encoding mode parameters; a transmission unit, configured to send the similarity information between the spectral signal of the audio signal in the current frequency band obtained by the judging unit and the spectral signal of the audio signal in the previous frequency band, and transmit the time-domain packet of the audio signal extracted by the encoding unit Envelope parameters and frequency domain envelope parameters are encoded and then sent; or, sending the information that the spectrum signal of the audio signal of the current frequency band obtained by the judgment unit is not similar to the spectrum signal between the audio signal of the previous frequency band, and the coded The time-domain envelope parameters, frequency-domain envelope parameters, pitch parameters and harmonic interval parameters of the audio signal extracted by the unit are encoded and then sent. 13. The encoding processing device according to claim 12, wherein: The encoding unit is configured to extract time-domain envelope parameters, frequency-domain envelope parameters, and When pitch parameters and harmonic interval parameters are used, it is also used to: When the harmonic interval of the audio signal is different from the value of the first harmonic offset, extracting the first harmonic offset parameter of the audio signal; The transmission unit is further configured to encode the first harmonic offset parameter and send it. 14. The encoding processing device according to claim 12, wherein: When the encoding unit performs the extraction of time-domain envelope parameters and frequency-domain envelope parameters used to characterize the audio signal when the spectral signal of the audio signal in the current frequency band is similar to the spectral signal of the audio signal in the previous frequency band, Also used for: extracting tone parameters of the audio signal; The transmission unit is further configured to encode the tone parameter and send it. 15. A decoding processing method, characterized in that, comprising: Receive the data sent by the encoding end, if the information indicating that the spectral signal of the audio signal in the current frequency band is similar to the spectral signal of the audio signal in the previous frequency band is received, according to the time domain envelope parameters and frequency domain envelope used to characterize the audio signal parametrically synthesized audio signals, wherein the time-domain envelope parameters and frequency-domain envelope parameters are decoded from received data; If information indicating that the spectral signal of the audio signal of the current frequency band is not similar to the spectral signal of the audio signal of the previous frequency band is received, according to the time domain envelope parameters, frequency domain envelope parameters, tone parameters and Synthesizing an audio signal with harmonic interval parameters, wherein the time-domain envelope parameter, frequency-domain envelope parameter, pitch parameter and harmonic interval parameter are decoded from received data, and the pitch parameter represents the harmonic signal in the audio signal The ratio to the noise signal; the information indicating whether the spectral signal of the audio signal in the current frequency band is similar or not similar to the spectral signal of the audio signal in the previous frequency band is specifically represented by a coding mode parameter. 16. The method according to claim 15, wherein the information indicating that the spectral signal of the audio signal of the current frequency band is similar or dissimilar to the spectral signal of the audio signal of the previous frequency band is specifically represented by a coding mode parameter, so The methods described include: According to the received coding mode parameter, it is determined whether the spectral signal of the audio signal of the current frequency band is similar or not similar to the spectral signal of the audio signal of the previous frequency band. 17. The method according to claim 16, wherein if the spectral signal of the audio signal of the current frequency band is similar to the spectral signal between the audio signal of the previous frequency band, then the step of synthesizing the audio signal comprises: Reconstructing the spectrum signal; processing the reconstructed spectrum signal according to the frequency domain envelope parameter and the time domain envelope parameter to obtain a synthesized audio signal. 18. The method according to claim 17, wherein the reconstructed spectral signal comprises: The spectrum signal is reconstructed by means of spectrum replication. 19. The method according to claim 17, wherein the data sent by the encoding end further comprises: a tone parameter for characterizing the audio signal; After the reconstructed spectral signal, it also includes: According to the tone parameter, the reconstructed spectrum signal is shaped to obtain the shaped reconstructed spectrum signal. 20. The method according to claim 15, wherein the data sent by the encoding end further includes: a first harmonic offset parameter. 21. The method according to claim 20, wherein if the spectral signal of the audio signal of the current frequency band is not similar to the spectral signal of the audio signal of the previous frequency band, then the step of synthesizing the audio signal comprises: obtaining a harmonic signal according to the harmonic interval parameter; or obtaining a harmonic signal according to the harmonic interval parameter and the first harmonic offset parameter; adjusting the ratio between the harmonic signal and the noise signal according to the tone parameter; and obtaining a reconstructed spectrum signal according to the adjusted harmonic signal and noise signal; Processing the reconstructed spectrum signal according to the frequency domain envelope parameter and the time domain envelope parameter to obtain a synthesized audio signal. 22. The method according to claim 17 or 21, wherein said processing the reconstructed spectrum signal according to said frequency domain envelope parameter and time domain envelope parameter to obtain a synthetic audio signal comprises: performing frequency domain shaping processing on the reconstructed spectral signal according to the frequency domain envelope parameters to obtain a frequency domain shaped signal, and performing time domain shaping processing on the frequency domain shaped signal according to the time domain envelope parameters, obtain a synthetic audio signal; or, performing time domain shaping processing on the reconstructed spectrum signal according to the time domain envelope parameters to obtain a time domain shaped signal, and performing frequency domain shaping processing on the time domain shaped signal according to the frequency domain envelope parameters, A synthesized audio signal is obtained. 23. A decoding processing device, characterized in that it comprises: The receiving information unit is used to receive information indicating that the spectral signal of the audio signal in the current frequency band is similar to the spectral signal of the audio signal in the previous frequency band, and decode the received data to obtain the time-domain envelope parameters and Frequency-domain envelope parameters; or, receiving information indicating that the spectral signal of the audio signal in the current frequency band is not similar to the spectral signal of the audio signal in the previous frequency band, and decoding the received data to obtain a time-domain packet for representing the audio signal An envelope parameter, a frequency domain envelope parameter, a tone parameter and a harmonic interval parameter, the tone parameter characterizes the ratio between the harmonic signal and the noise signal in the audio signal; Information about the similarity or dissimilarity of the spectral signal of the audio signal, specifically expressed by the encoding mode parameter; a decoding unit, configured to synthesize an audio signal according to the similar information received by the receiving information unit, and the time-domain envelope parameter and the frequency-domain envelope parameter; or, according to the dissimilar information, and the The time domain envelope parameter, the frequency domain envelope parameter, the tone parameter and the harmonic interval parameter are described to synthesize the audio signal. 24. The device according to claim 23, wherein when receiving information that the spectral signal of the audio signal in the current frequency band is similar to the spectral signal of the audio signal in the previous frequency band, the decoding unit specifically includes: Reconstruction subunit: used to reconstruct spectral signals; The first shaping subunit is configured to process the reconstructed spectrum signal according to the frequency domain envelope parameter and the time domain envelope parameter to obtain a synthesized audio signal. 25. The device according to claim 24, wherein the receiving information unit is further configured to decode the received data containing tone parameters, and obtain tone parameters used to characterize the audio signal; The decoding unit also includes: The second shaping subunit: for performing shaping processing on the spectral signal reconstructed by the reconstruction subunit according to the pitch parameter, to obtain the reconstructed spectral signal after shaping; The first shaping subunit processes the reconstructed spectral signal after shaping by the second shaping subunit according to the frequency domain envelope parameter and the time domain envelope parameter to obtain the synthesized audio signal. 26. The device according to claim 23, wherein when receiving information that the spectral signal of the audio signal in the current frequency band is not similar to the spectral signal of the audio signal in the previous frequency band, the received information unit When the data is decoded to obtain the time domain envelope parameters, frequency domain envelope parameters, pitch parameters and harmonic interval parameters used to characterize the audio signal, the data containing the first harmonic offset parameter is also received and decoded to obtain the The first harmonic offset parameter of the audio signal. 27. The device according to claim 26, wherein the decoding unit comprises: The harmonic reconstruction subunit is used to obtain a harmonic signal according to the harmonic interval parameter; or obtain a harmonic signal according to the harmonic interval parameter and the first harmonic offset parameter; The spectral signal reconstruction subunit is used to adjust the ratio between the harmonic signal and the noise signal according to the tone parameter, and obtain a reconstructed spectral signal according to the adjusted harmonic signal and noise signal; The third shaping subunit is configured to process the reconstructed spectrum signal according to the frequency domain envelope parameter and the time domain envelope parameter to obtain a synthesized audio signal.

Images