CN115762556A - Audio data processing method and system - Google Patents

Abstract

The application discloses an audio data processing method and system, after input data are obtained, the input data are divided into a plurality of sections of data to be processed according to frequency, after gain processing is carried out, a first low-pass filter with cut-off frequency matched with the cut-off frequency of each section of data to be processed carries out filtering processing on each section of data, distortion and noise which are larger than the corresponding cut-off frequency in attenuation section output data can be guaranteed, the problem of harmonic noise caused in a frequency division processing process is solved, and user experience is improved.

Description

Audio data processing method and system

Technical Field

The present application relates to the field of audio processing, and in particular, to an audio data processing method and system.

Background

In the process of digital signal processing, there will be an application requirement that the digital signal is divided into a plurality of frequency bands to be respectively subjected to dynamic range control DRC processing, and the results after different frequency band processing are superposed, namely, multi-band dynamic range control MBDRC. Among them, DRC (Dynamic range control) can give different gains according to different input signal amplitudes to make the sound softer or louder.

However, in the frequency division processing of the multiband dynamic range control MBDRC, when the divided low-band signal is subjected to enhancement processing, a high-frequency component is easily introduced, so that harmonic noise of the low-band signal becomes large, the noise of the low-band signal becomes strong, and the hearing is affected.

Disclosure of Invention

In view of the above, the present application provides an audio data processing method and system, and the specific scheme is as follows:

an audio data processing method, comprising:

obtaining input data;

dividing the input data into data to be processed in different frequency bands according to the frequency;

determining gain data matched with each section of data to be processed, performing gain processing on the data to be processed matched with the gain data according to the gain data, and determining segmented data corresponding to each section of the data to be processed, wherein the gain data matched with the data to be processed in different frequency bands are different;

respectively determining first low-pass filters matched with each section of data to be processed, and performing filtering processing on the matched segmented data through each first low-pass filter to obtain segmented output data corresponding to each section of the segmented data, wherein the cut-off frequency of each section of data to be processed corresponds to the cut-off frequency of the matched first low-pass filter;

and obtaining output data corresponding to the input data based on the segmented output data.

Further, the cutoff frequency of each of the data to be processed corresponds to the cutoff frequency of the matched first low-pass filter, and the method includes:

and the cut-off frequency of each first low-pass filter corresponding to each section of data to be processed is a preset multiple of the cut-off frequency of each corresponding section of data to be processed.

Further, the determining gain data matched with each segment of data to be processed, performing gain processing on the data to be processed matched with the gain data according to the gain data, and determining segmented data corresponding to each segment of the data to be processed includes:

generating corresponding gain parameters based on each section of data to be processed;

smoothing the gain parameters, and filtering the smoothed gain parameters through a second low-pass filter to obtain gain data;

and multiplying each section of data to be processed by the matched gain data to obtain the segmented data corresponding to each section of data to be processed.

Further, the method also comprises the following steps:

and amplitude detection is carried out on each section of data to be processed to obtain an amplitude data signal, so that the gain parameter can be obtained by processing the amplitude data signal.

Further, the method also comprises the following steps:

down-sampling the amplitude data signal, and generating a gain parameter based on the down-sampled amplitude data signal;

and after smoothing the gain parameters, up-sampling the smoothed gain parameters.

Further, the amplitude detection on each section of data to be processed includes:

carrying out amplitude detection on each section of data to be processed in a peak detection mode;

or the like, or a combination thereof,

and carrying out amplitude detection on each section of data to be processed in a root mean square value detection mode.

Further, the dividing the input data into at least one section of data to be processed in different frequency bands according to the frequency size includes:

determining n frequency points in the input data, wherein n is a positive integer;

dividing the input data into n +1 pieces of data to be processed in different frequency bands according to n frequency points, wherein the maximum frequency of the data to be processed in the low frequency band in the data to be processed in the adjacent frequency bands is the minimum frequency of the data to be processed in the high frequency band.

An audio data processing system comprising:

the third filter bank is used for obtaining input data and dividing the input data into data to be processed in different frequency bands according to frequency;

the dynamic range control module is used for determining gain data matched with each section of data to be processed, performing gain processing on the data to be processed matched with the gain data according to the gain data, and determining segmented data corresponding to each section of the data to be processed, wherein the gain data matched with the data to be processed in different frequency bands are different;

the first low-pass filter bank is used for determining a first low-pass filter matched with each section of data to be processed, filtering the matched segmented data through each first low-pass filter to obtain segmented output data corresponding to each section of the segmented data, wherein the cut-off frequency of each section of data to be processed corresponds to the cut-off frequency of the matched first low-pass filter;

an output module to obtain output data corresponding to the input data based on the segmented output data.

Further, the dynamic range control module includes:

the gain generation module is used for generating corresponding gain parameters based on each section of data to be processed;

the gain smoothing module is used for smoothing the gain parameters;

the second low-pass filter is used for filtering the gain parameter after the smoothing treatment to obtain gain data;

and the computing module is used for multiplying the each section of data to be processed with the matched gain data to obtain the segmented data corresponding to the each section of data to be processed.

Further, the dynamic range control module further comprises:

the amplitude detection module is used for carrying out amplitude detection on each section of data to be processed to obtain an amplitude data signal;

a down-sampling module for down-sampling the amplitude magnitude data signal so as to generate a gain parameter based on the down-sampled amplitude magnitude data signal;

and the up-sampling module is used for up-sampling the gain parameters after smoothing processing.

According to the technical scheme, the audio data processing method and the system obtain input data, divide the input data into data to be processed in different frequency bands according to frequency, determine gain data matched with each segment of the data to be processed, perform gain processing on the data to be processed matched with the gain data according to the gain data, and determine segment data corresponding to each segment of the data to be processed, wherein the gain data matched with the data to be processed in different frequency bands are different; respectively determining first low-pass filters matched with each section of data to be processed, and performing filtering processing on the matched segmented data through each first low-pass filter to obtain segmented output data corresponding to each section of segmented data, wherein the cut-off frequency of each section of data to be processed corresponds to the cut-off frequency of the matched first low-pass filter; output data corresponding to the input data is obtained based on the segmented output data. According to the scheme, after the input data are obtained, the input data are divided into one or more sections of data to be processed according to the frequency, after gain processing is carried out, the first low-pass filter with the cut-off frequency matched with the cut-off frequency of each section of data to be processed is added to carry out filtering processing on each section of data, distortion and noise which are larger than the corresponding cut-off frequency in the attenuation section output data can be guaranteed, the problem of harmonic noise caused in the frequency division processing process is solved, and user experience is improved.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings used in the embodiments or the prior art descriptions will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative efforts.

Fig. 1 is a flowchart of an audio data processing method disclosed in an embodiment of the present application;

fig. 2 is a schematic diagram of a frequency division band of an input signal according to an embodiment of the present application;

fig. 3 is a schematic diagram of processing input data to obtain output data according to an embodiment of the present disclosure;

FIG. 4 is a flowchart of an audio data processing method disclosed in an embodiment of the present application;

fig. 5 is a schematic structural diagram of an audio data processing system according to an embodiment of the present application.

Detailed Description

The technical solutions in the embodiments of the present application will be described clearly and completely with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only some embodiments of the present application, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

The application discloses an audio data processing method, a flow chart of which is shown in fig. 1, comprising the following steps:

s11, obtaining input data;

s12, dividing input data into data to be processed in different frequency bands according to frequency;

step S13, determining gain data matched with each section of data to be processed, performing gain processing on the data to be processed matched with the gain data according to the gain data, and determining section data corresponding to each section of the data to be processed, wherein the gain data matched with the data to be processed in different frequency bands are different;

step S14, respectively determining first low-pass filters matched with each section of data to be processed, and performing filtering processing on the matched segmented data through each first low-pass filter to obtain segmented output data corresponding to each section of segmented data, wherein the cut-off frequency of each section of data to be processed corresponds to the cut-off frequency of the matched first low-pass filter;

and S15, obtaining output data corresponding to the input data based on the segmented output data.

In the process of processing the digital signal, the digital signal is divided into frequency bands to independently perform DRC (Dynamic range control) processing, and then the digital signal is superimposed.

In order to solve the problem, in the scheme, after the gain processing is performed on the low-frequency-band signal obtained by controlling the frequency division of the multi-band dynamic range, the first low-pass filter is used for adding the filtering processing to the low-frequency-band signal obtained by frequency division so as to reduce distortion and noise.

When a section of audio data is obtained and multi-section dynamic range control is to be performed on the audio data, when the audio data is divided into sections, the number of divided sections can be randomly set, can be determined based on user setting, and can be determined based on actual requirements of the audio data.

When the frequency band of the data is divided, the frequency may be divided based on the size of the frequency, specifically: determining n frequency points in input data, wherein n is a positive integer; dividing input data into n +1 pieces of data to be processed in different frequency bands according to n frequency points, wherein the maximum frequency of the data to be processed in the low frequency band in the data to be processed in the adjacent frequency bands is the minimum frequency of the data to be processed in the high frequency band.

That is, if the number of frequency points is predetermined, each frequency point is used as a boundary point of different frequency bands.

For example: inputting data Din, determining the number of frequency points to be 4, dividing the input data into 5 sections, and determining the frequency points to be Fc1, fc2, fc3 and Fc4, wherein the Fc1, the Fc2, the Fc3 and the Fc4 are arranged in the order from small to large. The input data Din is divided into 5 sections of data to be processed, namely, the frequency band of the first section of data to be processed is 0-Fc1, the frequency band of the second section of data to be processed is Fc1-Fc2, the frequency band of the third section of data to be processed is Fc2-Fc3, the frequency band of the fourth section of data to be processed is Fc3-Fc4, and the frequency band of the fifth section of data to be processed is the highest frequency in Fc 4-input data.

In addition, the number of frequency bands may be determined first, and then the frequency bands may be divided in a frequency-average allocation manner.

The division of different frequency bands is carried out on the input data, and the division can be realized through a third filter bank. Both low-pass and high-pass filters are included in the third filter bank, wherein the cut-off frequency of each filter is the same as one of the frequency points.

Specifically, a first section of data to be processed with the lowest frequency band is obtained through a first low-pass filter LPF1 in a third filter bank, and the cutoff frequency of the first low-pass filter LPF1 in the third filter bank is the cutoff frequency in the first section of data to be processed, that is, fc1; obtaining the data to be processed with the highest frequency band through the last high-pass filter in the third filter bank, wherein the cut-off frequency of the high-pass filter is the lowest frequency in the data to be processed; the data to be processed except the first section and the last section of the data to be processed needs to be filtered by a low-pass filter and a high-pass filter at the same time, so as to obtain the data to be processed with the frequency band between the cut-off frequency of the low-pass filter and the cut-off frequency of the high-pass filter.

Continuing with the above example, the third filter bank includes: a first low pass filter LPF1, a second low pass filter LPF2, a third low pass filter LPF3, a fourth low pass filter LPF4, a first high pass filter HPF1, a second high pass filter HPF2, a third high pass filter HPF3 and a fourth high pass filter HPF4. Wherein, the cutoff frequency of the LPF1 and the HPF1 is Fc1, the cutoff frequency of the LPF2 and the HPF2 is Fc2, the cutoff frequency of the LPF3 and the HPF3 is Fc3, and the cutoff frequency of the LPF4 and the HPF4 is Fc4.

The input signal Din passes through an LPF1, so that a low-frequency signal Din1 lower than Fc1 can be obtained; the input signal Din passes through an LPF2 and an HPF1, so that a frequency range signal Din2 higher than Fc1 and lower than Fc2 can be obtained; the input signal Din passes through an LPF3 and an HPF2, so that a frequency band signal Din3 higher than Fc2 and lower than Fc3 can be obtained; the input signal Din passes through an LPF4 and an HPF3, so that a frequency range signal Din4 higher than Fc3 and lower than Fc4 can be obtained; the input signal Din passes through the HPF4, so that a high-frequency signal Din5 higher than Fc4 can be obtained.

As shown in FIG. 2, the process from Din to Din1-Din5 is the frequency division process of the MBDRC multi-band dynamic range control, which divides the input signal Din into a plurality of different frequency bands according to the actual sound effect processing requirement.

Of course, the specific frequency band may be divided in a plurality of ways, and is not limited in detail herein.

After the signal division is completed, at least one section of data to be processed can be obtained, such as: din1-Din5, respectively performing subsequent processing steps on each section of data to be processed in at least one section of data to be processed, namely performing gain processing on each section of data to be processed based on matched gain data to obtain corresponding segmented data, performing filtering processing on the section of data to be processed through a first low-pass filter matched with the section of data to be processed to obtain corresponding segmented output data, and finally overlapping at least one section of segmented output data to obtain complete output data corresponding to input data.

That is, there is a corresponding set of processing modules for each section of data to be processed, such as: the dynamic range control module and the first low-pass filter are different in control parameters in the dynamic range control module corresponding to different data to be processed, and different in filtering parameters of the first low-pass filter corresponding to different data to be processed.

The data to be processed is subjected to Gain processing through the dynamic range control module, the Gain processing is dynamic time-varying Gain generated in real time according to the signal amplitude of the data to be processed, and due to the fact that the signal amplitude of the data to be processed is randomly varied, any frequency component exists on the Gain.

And determining the data after gain processing as segmented data, and then performing filtering processing on the segmented data through a first low-pass filter, wherein the cut-off frequency of the first low-pass filter is matched with the cut-off frequency of the segment of data to be processed so as to attenuate signals larger than the cut-off frequency of the segment of data to be processed to ensure the authenticity of sound.

The cut-off frequency of the first low-pass filter is greater than the cut-off frequency of the section of data to be processed, so that signals greater than the cut-off frequency of the data to be processed can be filtered.

For example: the cut-off frequency of the first low-pass filter is a preset multiple of the cut-off frequency of the section of data to be processed, such as: 2 times, 3 times, etc.

Because the frequency response of the filter has a transition band from no attenuation to attenuation, the cut-off frequency of the first low-pass filter is set to be 2 times of the cut-off frequency of the section of data to be processed, and the influence of the first low-pass filter on signals in the section of data, which are smaller than the cut-off frequency of the section of data to be processed, can be avoided.

Or, the specific value of the preset multiple relationship may also be set according to the actual specific transition band of the first low-pass filter, the value of the preset multiple may be adjusted to be larger for the first low-pass filter with a wider transition band, and the relationship of the preset multiple may be adjusted to be smaller for the first low-pass filter with a narrower transition band.

After each section of data to be processed is subjected to gain and filtering processing, each section can obtain corresponding segmented output data, and all the segmented output data are overlapped to obtain complete output data corresponding to the input data.

As shown in fig. 3, the input data Din is divided into 5 segments of data Din1-Din5 to be processed, din1 is subjected to gain processing by the first dynamic range control module DRC1 to obtain first segment data Dout1, and the first segment data Dout1 is subjected to gain processing by the first low-pass filter LPF12 to obtain first segment output data; din2 is subjected to gain processing of a second dynamic range control module DRC2 to obtain second sectional data Dout2, and the second sectional data Dout2 is subjected to second low pass filter LPF22 to obtain second sectional output data; din3 is subjected to gain processing of a third dynamic range control module DRC3 to obtain third segment data Dout3, and the third segment data Dout3 is subjected to gain processing of a third first low-pass filter LPF32 to obtain third segment output data; din4 is subjected to gain processing of a fourth dynamic range control module DRC4 to obtain fourth subsection data Dout4, and the fourth subsection data Dout4 is subjected to fourth low pass filter LPF42 to obtain fourth subsection output data; din5 is subjected to gain processing by a fifth dynamic range control module DRC5 to obtain fifth segment data Dout5, and the fifth segment data Dout5 is subjected to a fifth low-pass filter LPF52 to obtain fifth segment output data; and superposing the first segment output data, the second segment output data, the third segment output data, the fourth segment output data and the fifth segment output data to obtain complete output data Dout corresponding to the input data Din.

Further, if the input data can be divided into only one segment of data to be processed according to the frequency, the data to be processed is processed based on the above method, and then a segment of segmented output data corresponding to the segment of data to be processed is obtained.

The audio data processing method disclosed in this embodiment obtains input data, divides the input data into to-be-processed data in different frequency bands according to frequency, determines gain data matched with each segment of the to-be-processed data, performs gain processing on the to-be-processed data matched with the gain data according to the gain data, determines segment data corresponding to each segment of the to-be-processed data, and determines that the gain data matched with the to-be-processed data in different frequency bands are different; respectively determining first low-pass filters matched with each section of data to be processed, and performing filtering processing on the matched segmented data through each first low-pass filter to obtain segmented output data corresponding to each section of segmented data, wherein the cut-off frequency of each section of data to be processed corresponds to the cut-off frequency of the matched first low-pass filter; output data corresponding to the input data is obtained based on the segmented output data. According to the scheme, after the input data are obtained, the input data are divided into one or more sections of data to be processed according to the frequency, after gain processing is carried out, the first low-pass filter with the cut-off frequency matched with the cut-off frequency of each section of data to be processed is added to carry out filtering processing on each section of data, distortion and noise which are larger than the corresponding cut-off frequency in the attenuation section output data can be guaranteed, the problem of harmonic noise caused in the frequency division processing process is solved, and user experience is improved.

The present embodiment discloses an audio data processing method, a flowchart of which is shown in fig. 4, and includes:

s41, obtaining input data;

step S42, dividing the input data into at least sections of data to be processed in different frequency bands according to the frequency;

s43, generating corresponding gain parameters based on each section of data to be processed;

step S44, smoothing the gain parameter, and filtering the smoothed gain parameter through a second low-pass filter to obtain gain data;

s45, multiplying each section of data to be processed by the matched gain data to obtain segmented data corresponding to each section of data to be processed;

step S46, respectively determining first low-pass filters matched with each section of data to be processed, and performing filtering processing on the matched segmented data through each first low-pass filter to obtain segmented output data corresponding to each section of segmented data, wherein the cut-off frequency of each section of data to be processed corresponds to the cut-off frequency of the matched first low-pass filter;

and S47, obtaining output data corresponding to the input data based on the segmented output data.

In the process of performing gain processing on each segment of data to be processed, a gain parameter needs to be generated by a gain generation module GainComputer first, and then the gain parameter is smoothed by a gain smoothing module GainSmooth, so that when the processed gain data is multiplied by the corresponding data to be processed, the required segmented data can be obtained.

A plurality of Gain parameters can be obtained based on a piece of data to be processed, and a large sudden change may exist between two consecutive Gain parameter Gain1 values obtained by the Gain generation module, and if the sudden change directly acts on input data, a signal output after DRC adjustment by the dynamic range control module has a sudden change in many scenarios, such as: in the audio data processing scenario, it is not acceptable, and therefore, a gain smoothing module is required to smooth abrupt changes of the output data.

Wherein, the gain generating module GainComputer has a Threshold value Threshold, a turning smooth Knee and a slope of DRC curve, wherein, the Threshold value Threshold represents a Threshold value for triggering the compression function of the Compressor, and when the Threshold value is exceeded, the Compressor starts to suppress the input signal; ratio represents the slope of the suppressed input signal, and the larger the Ratio is, the more the suppression is; knee represents a smooth transition in the process of the signal being un-suppressed and being suppressed, ensuring the effect of the processed signal when heard.

Further, the audio data processing method disclosed in this embodiment may further include:

amplitude detection is carried out on each section of data to be processed to obtain an amplitude data signal, so that gain parameters can be obtained through processing the amplitude data signal.

That is, the gain parameter is actually amplitude-related data obtained by detecting the amplitude of the data to be processed. The module for detecting the amplitude value may be a peak detection module PeakDet, or may be a root mean square value detection module RmsDet. Of course, other detection modules may also be used, and are not limited in this respect.

After obtaining the amplitude-related data through amplitude detection, the following steps may be further performed: and performing down-sampling on the amplitude data signal, and generating a gain parameter based on the down-sampled amplitude data signal.

And after the amplitude detection is carried out, the amplitude Data signal Data1 is obtained, the amplitude Data signal Data1 enters a down-sampling DS module, and partial Data in the Data1 is extracted through the down-sampling DS module so as to reduce the number of the Data. The extraction method can be as follows: and (3) extracting every m points at equal intervals, or taking the maximum value of every m points to perform down-sampling and the like, wherein m can be set according to actual requirements.

And the data processing amount of the gain generation module and the gain smoothing module is reduced through the down-sampling of the down-sampling module.

After down-sampling is performed by the down-sampling module, gain parameters are generated and smoothed, and then up-sampling is performed by the up-sampling module, so that the sampling rate of gain data which can be multiplied by the data to be processed can be matched with the data to be processed.

There are a compression time AttackTime and a release time ReleaseTime in the gain smoothing module. Because the time length of the compression time and the release time is far longer than the time interval Ts of the sampling points of the signal input to the dynamic range control module, the Gain updating of each sampling point in the input signal is not needed, and the Gain is updated only in a period of time and then is subjected to smoothing processing.

Among them, the up-sampling method includes linear interpolation, 0-order hold, spline interpolation, and the like.

After the gain parameter is smoothed, filtering is performed through a second low-pass filter to obtain gain data, so that the finally obtained gain data is multiplied by corresponding data to be processed, and thus segmented data corresponding to each segment of data to be processed is obtained.

The cut-off frequency of the second low-pass filter can be set according to the actual application effect or requirement, if the cut-off frequency is set to be smaller, the attenuation of the high-frequency component is more, and the harmonic noise of the output segmented data is smaller; if the cutoff frequency is set too small, the gain parameter after the smoothing process is changed slowly, and the size of the output segment data is not adjusted to a predetermined value.

It should be noted that each segment of data to be processed corresponds to one dynamic range control module, and each dynamic range control module at least includes: the gain generating module, the gain smoothing module, the second low-pass filter and the calculating module, in addition, each dynamic range control module may further include: the device comprises an amplitude detection module, a down-sampling module and an up-sampling module.

The audio data processing method disclosed by the embodiment includes the steps of obtaining input data, dividing the input data into data to be processed in different frequency bands according to frequency, determining gain data matched with each segment of the data to be processed, performing gain processing on the data to be processed matched with the gain data according to the gain data, and determining segment data corresponding to each segment of the data to be processed, wherein the gain data matched with the data to be processed in different frequency bands are different; respectively determining first low-pass filters matched with each section of data to be processed, and performing filtering processing on the matched segmented data through each first low-pass filter to obtain segmented output data corresponding to each section of segmented data, wherein the cut-off frequency of each section of data to be processed corresponds to the cut-off frequency of the matched first low-pass filter; and obtaining output data corresponding to the input data based on the segmented output data. According to the scheme, after the input data are obtained, the input data are divided into one or more sections of data to be processed according to the frequency, after gain processing is carried out, the first low-pass filter with the cut-off frequency matched with the cut-off frequency of each section of data to be processed is added to carry out filtering processing on each section of data, distortion and noise which are larger than the corresponding cut-off frequency in the attenuation section output data can be guaranteed, the problem of harmonic noise caused in the frequency division processing process is solved, and user experience is improved.

The present embodiment discloses an audio data processing system, a schematic structural diagram of which is shown in fig. 5, and the audio data processing system includes:

a third filter bank 51, a dynamic range control module 52, a first low pass filter bank 53 and an output module 54.

The third filter bank 51 is configured to obtain input data, and divide the input data into data to be processed in different frequency bands according to frequency;

the dynamic range control module 52 is configured to determine gain data matched with each segment of data to be processed, perform gain processing on the data to be processed matched with the gain data according to the gain data, and determine segment data corresponding to each segment of data to be processed, where the gain data matched with the data to be processed in different frequency bands are different;

the first low-pass filter bank 53 is configured to determine a first low-pass filter matched with each segment of data to be processed, and perform filtering processing on the matched segment data through each first low-pass filter to obtain segment output data corresponding to each segment of segment data, where a cutoff frequency of each piece of data to be processed corresponds to a cutoff frequency of the matched first low-pass filter;

the output module 54 is configured to obtain output data corresponding to the input data based on the segmented output data.

In the process of processing the digital signal, the digital signal is divided into frequency bands to independently perform DRC (Dynamic range control) processing, and then the digital signal is superimposed.

In order to solve the problem, in the scheme, after the gain processing is performed on the low-frequency-band signal obtained by controlling the frequency division of the multi-band dynamic range, the first low-pass filter is used for adding the filtering processing to the low-frequency-band signal obtained by frequency division so as to reduce distortion and noise.

When a piece of audio data is obtained and multi-segment dynamic range control is to be performed on the audio data, the number of divided frequency segments can be randomly set when the audio data is divided, can be determined based on user setting, and can be determined based on actual requirements of the audio data.

When the frequency band of the data is divided, the data may be divided based on the size of the frequency, specifically: determining n frequency points in input data, wherein n is a positive integer; dividing input data into n +1 pieces of data to be processed in different frequency bands according to n frequency points, wherein the maximum frequency of the data to be processed in the low frequency band in the data to be processed in the adjacent frequency bands is the minimum frequency of the data to be processed in the high frequency band.

That is, if the number of frequency points is predetermined, each frequency point is used as a boundary point of different frequency bands.

For example: inputting data Din, determining the number of frequency points to be 4, dividing the input data into 5 sections, and determining the frequency points to be Fc1, fc2, fc3 and Fc4, wherein the Fc1, the Fc2, the Fc3 and the Fc4 are arranged in the order from small to large. The input data Din is divided into 5 sections of data to be processed, namely, the frequency band of the first section of data to be processed is 0-Fc1, the frequency band of the second section of data to be processed is Fc1-Fc2, the frequency band of the third section of data to be processed is Fc2-Fc3, the frequency band of the fourth section of data to be processed is Fc3-Fc4, and the frequency band of the fifth section of data to be processed is the highest frequency in Fc 4-input data.

Alternatively, the number of frequency bands may be determined first, and then the frequency bands may be divided in a frequency-average allocation manner.

The division of different frequency bands is carried out on the input data, and the division can be realized through a third filter bank. Both low-pass and high-pass filters are included in the third filter bank, wherein the cut-off frequency of each filter is the same as one of the frequency bins.

Specifically, a first section of data to be processed with the lowest frequency band is obtained through a first low-pass filter LPF1 in a third filter bank, and the cutoff frequency of the first low-pass filter LPF1 in the third filter bank is the cutoff frequency in the first section of data to be processed, that is, fc1; obtaining the data to be processed with the highest frequency band through the last high-pass filter in the third filter bank, wherein the cut-off frequency of the high-pass filter is the lowest frequency in the data to be processed; except for the data to be processed of the first section and the last section in the data to be processed, the data to be processed needs to be filtered by the low-pass filter and the high-pass filter at the same time, so that the data to be processed with the frequency band between the cutoff frequency of the low-pass filter and the cutoff frequency of the high-pass filter is obtained.

Continuing with the above example, the third filter bank includes: a first low pass filter LPF1, a second low pass filter LPF2, a third low pass filter LPF3, a fourth low pass filter LPF4, a first high pass filter HPF1, a second high pass filter HPF2, a third high pass filter HPF3 and a fourth high pass filter HPF4. Wherein, the cutoff frequency of the LPF1 and the HPF1 is Fc1, the cutoff frequency of the LPF2 and the HPF2 is Fc2, the cutoff frequency of the LPF3 and the HPF3 is Fc3, and the cutoff frequency of the LPF4 and the HPF4 is Fc4.

The input signal Din passes through an LPF1, so that a low-frequency signal Din1 lower than Fc1 can be obtained; the input signal Din passes through an LPF2 and an HPF1, so that a frequency range signal Din2 higher than Fc1 and lower than Fc2 can be obtained; the input signal Din passes through an LPF3 and an HPF2, so that a frequency band signal Din3 higher than Fc2 and lower than Fc3 can be obtained; the input signal Din passes through an LPF4 and an HPF3, so that a frequency band signal Din4 higher than Fc3 and lower than Fc4 can be obtained; the input signal Din passes through the HPF4, so that a high-frequency signal Din5 higher than Fc4 can be obtained.

As shown in FIG. 2, the process from Din to Din1-Din5 is the frequency division process of the MBDRC multi-band dynamic range control, which divides the input signal Din into a plurality of different frequency bands according to the actual sound effect processing requirement.

Of course, the specific frequency band may be divided in a plurality of ways, and is not limited in detail herein.

After the signal division is completed, at least one section of data to be processed can be obtained, such as: din1-Din5, respectively performing subsequent processing steps on each section of data to be processed in at least one section of data to be processed, namely performing gain processing on each section of data to be processed based on matched gain data to obtain corresponding segmented data, performing filtering processing on the section of data to be processed through a first low-pass filter matched with the section of data to be processed to obtain corresponding segmented output data, and finally overlapping at least one section of segmented output data to obtain complete output data corresponding to input data.

That is, there is a corresponding set of processing modules for each section of data to be processed, such as: the dynamic range control module and the first low-pass filter are different in control parameters in the dynamic range control module corresponding to different data to be processed, and different in filtering parameters of the first low-pass filter corresponding to different data to be processed.

The data to be processed is subjected to Gain processing through the dynamic range control module, the Gain processing is dynamic time-varying Gain generated in real time according to the signal amplitude of the data to be processed, and due to the fact that the signal amplitude of the data to be processed is randomly varied, any frequency component exists on the Gain.

And determining the data after gain processing as segmented data, and then performing filtering processing on the segmented data through a first low-pass filter, wherein the cut-off frequency of the first low-pass filter is matched with the cut-off frequency of the segment of data to be processed so as to attenuate signals larger than the cut-off frequency of the segment of data to be processed to ensure the authenticity of sound.

The cut-off frequency of the first low-pass filter is greater than the cut-off frequency of the section of data to be processed, so that signals greater than the cut-off frequency of the data to be processed can be filtered.

For example: the cut-off frequency of the first low-pass filter is a preset multiple of the cut-off frequency of the section of data to be processed, such as: 2 times, 3 times, etc.

Because the frequency response of the filter has a transition band from no attenuation to attenuation, the cut-off frequency of the first low-pass filter is set to be 2 times of the cut-off frequency of the section of data to be processed, and the influence of the first low-pass filter on signals in the section of data, which are smaller than the cut-off frequency of the section of data to be processed, can be avoided.

Or, the specific value of the preset multiple relationship may also be set according to the actual specific transition band of the first low-pass filter, the value of the preset multiple may be adjusted to be larger for the first low-pass filter with a wider transition band, and the relationship of the preset multiple may be adjusted to be smaller for the first low-pass filter with a narrower transition band.

After each section of data to be processed is subjected to gain and filtering processing, each section can obtain corresponding segmented output data, and all the segmented output data are overlapped to obtain complete output data corresponding to the input data.

As shown in fig. 3, the input data Din is divided into 5 segments of data Din1-Din5 to be processed, din1 is subjected to gain processing by the first dynamic range control module DRC1 to obtain first segment data Dout1, and the first segment data Dout1 is subjected to gain processing by the first low-pass filter LPF12 to obtain first segment output data; din2 is subjected to gain processing by a second dynamic range control module DRC2 to obtain second segmented data Dout2, and the second segmented data Dout2 is subjected to second low pass filter LPF22 to obtain second segmented output data; din3 is subjected to gain processing by a third dynamic range control module DRC3 to obtain third segment data Dout3, and the third segment data Dout3 is subjected to third low pass filter LPF32 to obtain third segment output data; din4 is subjected to gain processing of a fourth dynamic range control module DRC4 to obtain fourth segment data Dout4, and the fourth segment data Dout4 is subjected to fourth low pass filter LPF42 to obtain fourth segment output data; din5 is subjected to gain processing of a fifth dynamic range control module DRC5 to obtain fifth-segment data Dout5, and the fifth-segment data Dout5 is subjected to a fifth low-pass filter LPF52 to obtain fifth-segment output data; and superposing the first segment output data, the second segment output data, the third segment output data, the fourth segment output data and the fifth segment output data to obtain complete output data Dout corresponding to the input data Din.

Further, the dynamic range control module includes: the gain generating module, the gain smoothing module, the second low-pass filter and the calculating module.

The gain generating module is used for generating corresponding gain parameters based on each section of data to be processed;

the gain smoothing module is used for smoothing the gain parameters;

the second low-pass filter is used for filtering the gain parameters subjected to the smoothing processing to obtain gain data;

and the computing module is used for multiplying each section of data to be processed by the matched gain data to obtain the subsection data corresponding to each section of data to be processed.

In the process of performing gain processing on each segment of data to be processed, a gain parameter needs to be generated by a gain generation module GainComputer, and then the gain parameter is smoothed by a gain smoothing module GainSmooth, so that when the processed gain data is multiplied by corresponding data to be processed, segmented data meeting requirements can be obtained.

A plurality of Gain parameters can be obtained based on a piece of data to be processed, and a large sudden change may exist between two consecutive Gain parameter Gain1 values obtained by the Gain generation module, and if the sudden change directly acts on input data, a signal output after DRC adjustment by the dynamic range control module has a sudden change in many scenarios, such as: in the audio data processing scenario, it is not acceptable, and therefore, a gain smoothing module is required to smooth abrupt changes of the output data.

Wherein, the gain generating module GainComputer has a Threshold value Threshold, a turning smooth Knee and a slope of DRC curve, wherein, the Threshold value Threshold represents a Threshold value triggering the compression function of the Compressor, and when the Threshold value is exceeded, the Compressor starts to suppress the input signal; the Ratio represents the slope of the suppressed input signal, and the larger the Ratio is, the more the suppression is; knee represents a smooth transition in the process of the signal being un-suppressed and being suppressed, ensuring the effect of the processed signal when heard.

Further, the audio data processing system disclosed in this embodiment may be further configured to:

amplitude detection is carried out on each section of data to be processed to obtain an amplitude data signal, so that gain parameters can be obtained through processing the amplitude data signal.

That is, the gain parameter is actually amplitude-related data obtained by detecting the amplitude of the data to be processed. The module for detecting the amplitude value may be a peak detection module PeakDet, or may be a root mean square value detection module RmsDet. Of course, other detection modules may be used, and are not limited in particular.

After obtaining the amplitude-related data through amplitude detection, the following steps may be performed: and performing down-sampling on the amplitude data signal, and generating a gain parameter based on the down-sampled amplitude data signal.

And after the amplitude detection is carried out, the amplitude Data signal Data1 is obtained, the amplitude Data signal Data1 enters a down-sampling DS module, and partial Data in the Data1 is extracted through the down-sampling DS module so as to reduce the number of the Data. The extraction method can be as follows: and extracting every m points at equal intervals, or taking the maximum value of every m points to perform down-sampling and the like, wherein m can be set according to actual requirements.

And the data processing amount of the gain generation module and the gain smoothing module is reduced through the down-sampling of the down-sampling module.

After down-sampling is performed by the down-sampling module, gain parameters are generated and smoothed, and then up-sampling is performed by the up-sampling module, so that the sampling rate of gain data which can be multiplied by the data to be processed can be matched with the data to be processed.

There are a compression time AttackTime and a release time ReleaseTime in the gain smoothing module. Because the time length of the compression time and the release time is far longer than the time interval Ts of the sampling points of the signal input to the dynamic range control module, the Gain updating of each sampling point in the input signal is not needed, and the Gain is updated only for a period of time and then is smoothed.

Among them, the up-sampling method includes linear interpolation, 0-order hold, spline interpolation, and the like.

After the gain parameter is smoothed, filtering is performed through a second low-pass filter to obtain gain data, so that the finally obtained gain data is multiplied by corresponding data to be processed, and thus segmented data corresponding to each segment of data to be processed is obtained.

The cut-off frequency of the second low-pass filter can be set according to the actual application effect or requirement, if the cut-off frequency is set to be smaller, the attenuation of the high-frequency component is more, and the harmonic noise of the output segmented data is smaller; if the cutoff frequency is set too small, the gain parameter after the smoothing process is changed slowly, and the size of the output segment data is not adjusted to a predetermined value.

It should be noted that each segment of data to be processed corresponds to one dynamic range control module, and each dynamic range control module at least includes: the gain generating module, the gain smoothing module, the second low-pass filter and the calculating module, in addition, each dynamic range control module may further include: the device comprises an amplitude detection module, a down-sampling module and an up-sampling module.

The audio data processing system disclosed in this embodiment obtains input data, divides the input data into to-be-processed data in different frequency bands according to frequency, determines gain data matched with each segment of the to-be-processed data, performs gain processing on the to-be-processed data matched with the gain data according to the gain data, determines segment data corresponding to each segment of the to-be-processed data, and determines that the gain data matched with the to-be-processed data in different frequency bands are different; respectively determining first low-pass filters matched with each section of data to be processed, and performing filtering processing on the matched segmented data through each first low-pass filter to obtain segmented output data corresponding to each section of segmented data, wherein the cut-off frequency of each section of data to be processed corresponds to the cut-off frequency of the matched first low-pass filter; output data corresponding to the input data is obtained based on the segmented output data. According to the scheme, after the input data are obtained, the input data are divided into one or more sections of data to be processed according to the frequency, after gain processing is carried out, the first low-pass filter with the cut-off frequency matched with the cut-off frequency of each section of data to be processed is added to carry out filtering processing on each section of data, distortion and noise which are larger than the corresponding cut-off frequency in the attenuation section output data can be guaranteed, the problem of harmonic noise caused in the frequency division processing process is solved, and user experience is improved.

The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. The device disclosed in the embodiment corresponds to the method disclosed in the embodiment, so that the description is simple, and the relevant points can be referred to the description of the method part.

Those of skill would further appreciate that the various illustrative components and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the components and steps of the various examples have been described above generally in terms of their functionality in order to clearly illustrate this interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.

The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module may reside in Random Access Memory (RAM), memory, read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims (10)

1. A method of audio data processing, comprising:

obtaining input data;

dividing the input data into data to be processed in different frequency bands according to the frequency;

determining gain data matched with each section of data to be processed, performing gain processing on the data to be processed matched with the gain data according to the gain data, and determining segmented data corresponding to each section of the data to be processed, wherein the gain data matched with the data to be processed in different frequency bands are different;

respectively determining first low-pass filters matched with each section of data to be processed, and performing filtering processing on the matched segmented data through each first low-pass filter to obtain segmented output data corresponding to each section of the segmented data, wherein the cut-off frequency of each section of data to be processed corresponds to the cut-off frequency of the matched first low-pass filter;

and obtaining output data corresponding to the input data based on the segmented output data.

2. The method of claim 1, wherein the cutoff frequency of each of the data to be processed corresponds to a cutoff frequency of a matched first low-pass filter, comprising:

and the cut-off frequency of each first low-pass filter corresponding to each section of data to be processed is a preset multiple of the cut-off frequency of each corresponding section of data to be processed.

3. The method of claim 1, wherein the determining gain data matching each segment of data to be processed, performing gain processing on the data to be processed matching the gain data according to the gain data, and determining segmented data corresponding to each segment of the data to be processed comprises:

generating corresponding gain parameters based on each section of data to be processed;

smoothing the gain parameters, and filtering the smoothed gain parameters through a second low-pass filter to obtain gain data;

and multiplying each section of data to be processed by the matched gain data to obtain the segmented data corresponding to each section of data to be processed.

4. The method of claim 3, further comprising:

and amplitude detection is carried out on each section of data to be processed to obtain an amplitude data signal, so that the gain parameter can be obtained by processing the amplitude data signal.

5. The method of claim 4, further comprising:

down-sampling the amplitude data signal, and generating a gain parameter based on the down-sampled amplitude data signal;

and after smoothing the gain parameters, up-sampling the smoothed gain parameters.

6. The method of claim 4, wherein the detecting the amplitude of each piece of data to be processed comprises:

carrying out amplitude detection on each section of data to be processed in a peak detection mode;

or the like, or, alternatively,

and carrying out amplitude detection on each section of data to be processed in a root mean square value detection mode.

7. The method of claim 1, wherein the dividing the input data into the data to be processed in different frequency bands according to the frequency size comprises:

determining n frequency points in the input data, wherein n is a positive integer;

dividing the input data into n +1 pieces of data to be processed in different frequency bands according to n frequency points, wherein the maximum frequency of the data to be processed in the low frequency band in the data to be processed in the adjacent frequency bands is the minimum frequency of the data to be processed in the high frequency band.

8. An audio data processing system, comprising:

the third filter bank is used for obtaining input data and dividing the input data into data to be processed in different frequency bands according to frequency;

the dynamic range control module is used for determining gain data matched with each section of data to be processed, performing gain processing on the data to be processed matched with the gain data according to the gain data, and determining segmented data corresponding to each section of the data to be processed, wherein the gain data matched with the data to be processed in different frequency bands are different;

the first low-pass filter bank is used for determining a first low-pass filter matched with each section of data to be processed, filtering the matched segmented data through each first low-pass filter to obtain segmented output data corresponding to each section of the segmented data, wherein the cut-off frequency of each section of data to be processed corresponds to the cut-off frequency of the matched first low-pass filter;

an output module to obtain output data corresponding to the input data based on the segmented output data.

9. The system of claim 8, wherein the dynamic range control module comprises:

the gain generation module is used for generating corresponding gain parameters based on each section of data to be processed;

the gain smoothing module is used for smoothing the gain parameters;

the second low-pass filter is used for filtering the gain parameter after the smoothing treatment to obtain gain data;

and the computing module is used for multiplying the each section of data to be processed with the matched gain data to obtain the segmented data corresponding to the each section of data to be processed.

10. The system of claim 9, wherein the dynamic range control module further comprises:

the amplitude detection module is used for carrying out amplitude detection on each section of data to be processed to obtain an amplitude data signal;

a down-sampling module for down-sampling the amplitude magnitude data signal so as to generate a gain parameter based on the down-sampled amplitude magnitude data signal;

and the up-sampling module is used for up-sampling the gain parameters after smoothing processing.

Images