CN114613375A - Time domain noise shaping method and device for audio signal - Google Patents

Abstract

The application provides a time domain noise shaping method and a time domain noise shaping device for an audio signal, and relates to the technical field of computers_k(ii) a k is any one of 1 to mAn integer number; m is the filtering order of the filter in the time domain noise shaping TNS; if the k-th prediction error value e_kGreater than the k-th order prediction error contrast value e _ thresh_kDirectly carrying out frequency spectrum quantization processing on the frequency domain signal; and if the prediction error values of all orders of the frequency domain signal are not larger than the corresponding prediction error comparison values of all orders, filtering the frequency domain signal through a filter in the TNS. The mode judges according to the comparison value of the prediction error value and the prediction error, and determines whether to execute the operation of filtering processing, thereby reducing the calculation amount and improving the data processing efficiency.

Description

A time-domain noise shaping method and device for audio signals

technical field

The present invention relates to the field of computer technology, and in particular, to a time-domain noise shaping method and device for audio signals.

Background technique

With the development of science and technology, low complexity communication codec (LC3) is widely used in audio equipment. However, big data tests show that in LC3, the standard temporal noise shaping (TNS) module takes up more than 16% of the entire encoding process, which is too high for the device to carry.

SUMMARY OF THE INVENTION

The present application provides a time-domain noise shaping method and apparatus for an audio signal, so as to reduce the calculation amount of time-domain noise processing.

In the first aspect, the present application provides a time-domain noise shaping method for audio signals. The method can be performed by an LC3 codec of an electronic device. The electronic device can be a Bluetooth headset, a mobile phone, etc., which is not specifically described in this application. limited. Execute as follows:

Determine the k-th order prediction error value e _k of the frequency domain signal processed by frequency domain noise shape (spectral noise shape, SNS); k is any integer from 1 to m; m is the filtering order of the filter in the TNS; If the k-th order prediction error value e _k is greater than the k-th order prediction error comparison value e_thresh _k , the frequency domain signal is subjected to spectral quantization; , the frequency domain signal is filtered through the filter in TNS.

It should be noted that, after receiving the audio signal in the present application, the electronic device can perform encoding processing on the audio signal to obtain a pulse code modulation (PCM) signal, and then perform Fourier transform on the PCM signal to convert the PCM signal into into a frequency domain signal for input into the LC3 codec for signal processing.

In this application, when the LC3 codec determines that the prediction error value of a certain order is greater than the prediction error comparison value, it does not perform filtering processing, and performs the operation of spectral quantization. In terms of the calculation of filtering processing, the calculation amount can be reduced, and the signal processing efficiency can obviously be improved when the calculation amount is reduced.

In an optional manner, for the frequency domain signal processed by SNS, determine the k-th order normalized autocorrelation coefficient r _w (k) of the frequency-domain signal, and k is initialized to 1; based on the k-th order normalization The autocorrelation coefficient r _w (k) determines the k-th order prediction error value e _k ; if the k-th order prediction error value e _k is not greater than the k-th order prediction error comparison value e_thresh _k , then assign the value of k plus 1 to k, return to the step of determining the k-th order normalized autocorrelation coefficient r _w (k) of the frequency domain signal, until the k-th order prediction error value e _k is greater than e_thresh _k or k is equal to m.

In the present application, by setting the loop operation, the accuracy of data processing can be guaranteed, and the loop can be ended when e _k is greater than the k-th order prediction error comparison value e_thresh _k or k=m, which can reduce the amount of calculation and further improve the data processing efficiency.

In an optional manner, for the frequency domain signal processed by the SNS, the normalized autocorrelation coefficient of each order of the frequency domain signal is determined; for any kth order, the normalization is based on the 1st to the k-1th order Calculate the autocorrelation coefficient to determine the k-th order prediction error value _ek of the frequency domain signal.

In this way, the prediction error value of each order can be directly determined, and the data processing efficiency is high.

In an optional manner, before the frequency domain signal is filtered by the filter in the TNS, based on the 0th order normalized autocorrelation coefficient r _w (0) and the _mth order prediction error value em , determine The predicted gain of the frequency domain signal; when the predicted gain is greater than the first preset value, the frequency domain signal is filtered by the filter in the TNS.

According to the relationship between the predicted gain and the first preset value, the present application determines whether to start the filtering operation, and in this way, the accuracy of the data processing can be ensured.

In an optional manner, when the prediction gain is not greater than the first preset value, spectral quantization processing is performed on the frequency domain signal.

In this way, the calculation amount of time-domain noise processing can be reduced, and the data processing efficiency can be further improved.

In an optional manner, any order prediction error comparison value in each order prediction error comparison value is determined by the corresponding order prediction error in the sample data; the sample data is a frequency domain signal that does not need to be processed by TNS filtering.

The prediction error comparison value determined in this way is more reliable, and the accuracy of data processing can be guaranteed.

In an optional manner, the comparison values of prediction errors of each order are determined under a set loss accuracy condition.

In this application, the comparison values of prediction errors of each order are related to the loss accuracy, and this method can ensure the accuracy of data processing.

In an optional manner, if each order prediction error value of the frequency domain signal is not greater than the corresponding comparison value of each order prediction error, the frequency domain signal is filtered by a filter in the TNS.

In this way, the accuracy of data processing can be guaranteed.

In an optional manner, the frequency domain signal is filtered by a filter in the TNS, including:

Based on the m-order linear prediction coding value of the frequency domain signal, the m-1 order reflection coefficient is determined; the m-1 order reflection coefficient is quantized to obtain the m-1 order quantized reflection coefficient of the filter in the TNS; based on the m-1 order reflection coefficient The first-order quantized reflection coefficient is used to filter the frequency domain signal through the filter in the TNS.

In a second aspect, the present application provides a time-domain noise shaping device for an audio signal, the device comprising:

An error determination unit, used for determining the k-th order prediction error value e _k of the frequency domain signal processed by the SNS; k is any integer from 1 to m; m is the filtering order of the filter in the time-domain noise shaping TNS; The first error comparison unit is used for performing spectrum quantization processing on the frequency domain signal if the kth order prediction error value e _k is greater than the kth order prediction error comparison value e_thresh _k ; the second error comparison unit is used for if the frequency domain signal is The prediction error value of each order is not greater than the corresponding comparison value of each order prediction error, then the frequency domain signal is filtered by the filter in the TNS.

In a third aspect, the present application provides a computing device, including: a memory and a processor; a memory for storing program instructions; a processor for calling program instructions stored in the memory, and executing the first aspect according to the obtained program the method described.

In a fourth aspect, the present application provides a computer storage medium storing computer-executable instructions for executing the method according to the first aspect.

For the technical effects that can be achieved by the second aspect to the fourth aspect, please refer to the description of the technical effects that can be achieved by the corresponding possible design solutions in the first aspect, which will not be repeated here in this application.

Other features and advantages of the present application will be set forth in the description which follows, and in part will be apparent from the description, or may be learned by practice of the present application. The objectives and other advantages of the application may be realized and attained by the structure particularly pointed out in the written description, claims, and drawings.

Description of drawings

In order to illustrate the technical solutions in the embodiments of the present invention more clearly, the following briefly introduces the accompanying drawings used in the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained from these drawings without any creative effort.

1 is a schematic diagram of an application scenario of a time-domain noise processing method for an audio signal;

2 is a schematic flowchart of a time-domain noise processing method;

FIG. 3 is a schematic flowchart of a time-domain noise shaping method provided by an embodiment of the present application;

4 is a schematic diagram of a prediction error comparison value provided by an embodiment of the present application;

FIG. 5 is a schematic flowchart of another time-domain noise shaping method provided by an embodiment of the present application;

FIG. 6 is a schematic structural diagram of a time-domain noise shaping apparatus provided by an embodiment of the present application;

FIG. 7 is a schematic structural diagram of a computing device according to an embodiment of the present application.

Detailed ways

In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present invention.

It should be noted that the terms "first", "second" and the like in this application are used to distinguish similar objects, and are not necessarily used to describe a specific sequence or sequence. It is to be understood that the data so used are interchangeable under appropriate circumstances such that the embodiments of the disclosure described herein can be practiced in sequences other than those illustrated or described herein. The implementations described in the illustrative examples below are not intended to represent all implementations consistent with this application. Rather, they are merely examples of apparatus and methods consistent with some aspects of the present application as recited in the appended claims.

In order to better illustrate the solution of this application, relative to the technologies that may be involved in this application, it is used to explain:

TNS (Time-Domain Noise Shaping): Time-domain noise shaping is to realize time-domain noise shaping by processing the signal spectrum, which can control the time-domain of quantization noise. The principle is to perform open-loop predictive coding on the time domain signal, so as to adjust the final quantization error power spectral density of the decoder to adapt to the power spectral density of the input signal.

LPC (Linear Predictive Coding): Approximate (predict) the current speech sample value by using the linear combination of several known past speech signals, so that the linear prediction value is equal to the current sample value of the speech at the minimum mean square error, and then The predicted value and the current sampled value are subtracted to obtain the prediction error, and finally the prediction error is encoded.

Prediction gain: used to judge whether to start TNS operation.

Prediction error: use the normalized windowed autocorrelation coefficient and prediction error to obtain the prediction gain, and determine whether to activate TNS by comparing the prediction gain and the set threshold (the first preset value in this application).

FIG. 1 shows the application scenario of the time-domain noise processing method for audio signals in this application. The application scenario includes: a device 1 for playing music (a Bluetooth speaker is used as an example in FIG. 1 ), and a Bluetooth headset, the Bluetooth speaker and the Audio data is transmitted between Bluetooth headsets through wireless communication. In this application scenario, Bluetooth communication is used. After the Bluetooth headset receives the audio data, the audio data is sampled, quantized, and encoded through PCM. Standard digital audio data, that is, PCM data. The LC3 codec is also set in the Bluetooth headset. The LC3 codec can convert the PCM data into a frequency domain signal, and input the frequency domain signal to the TNS module in the LC3 codec for noise filtering processing to obtain a filtered signal. The filtered signal can be decoded and converted into an audio signal for the user to listen to. In addition, it should be noted that, after the filtering process is performed, operations such as spectrum quantization are also performed, which are not described one by one in this application.

In the related art, the time-domain noise shaping process for audio signals is shown in Figure 2. Assuming that in practical application, the filtering order of the filter in the TNS is 8, it needs to go through 8-order (that is, 8 times) data processing, In order to better filter out the noise in the frequency domain signal, in Figure 2, the data processing starts from the acquired frequency domain signal. The LC3 codec can calculate the 8 normalized autocorrelation coefficients of the frequency domain signal respectively, based on the 8 normalized autocorrelation coefficients. The normalized autocorrelation coefficients calculate 8 LPC coefficients, and the prediction error is calculated based on the 8 LPC coefficients and the 8 normalized autocorrelation coefficients. Based on the normalized autocorrelation coefficient and the value of the prediction error, the prediction gain preGain is determined. When it is determined that the preGain is greater than 1.5, the reflection coefficient is calculated, the reflection coefficient is quantized, the TNS bit consumption is calculated, and the quantized reflection coefficient is used to filter the frequency domain signal to obtain a filtered signal. When it is determined that preGain is not greater than 1.5, the reflection coefficients are all set to 0, and quantization is performed based on the reflection coefficients of all 0s, the TNS bit consumption is calculated, and the quantized reflection coefficient is used to filter the frequency domain signal to obtain the filtered signal.

In this method, after the 8th-order data processing is all calculated uniformly, the subsequent time-domain noise shaping operation is performed based on the value of the prediction gain. Among them, the calculation of the standard time-domain noise shaping module accounts for more than 16% of the entire coding process, the calculation of the normalized autocorrelation coefficient, LPC coefficient and prediction error accounts for more than 80% of the calculation of the process, and the calculation and quantization of the reflection coefficient, The TNS bit consumption and the computation of filtering the spectrum using the quantized reflection coefficients account for more than 15% of the computation of the encoding process. This method requires a large amount of computation and is difficult for equipment to carry.

Therefore, reducing the calculation of normalized autocorrelation coefficients, LPC coefficients and prediction errors, as well as reducing the calculation and quantization of reflection coefficients, TNS bit consumption, and calculation of spectrum filtering using quantized reflection coefficients can effectively reduce the time-domain noise shaping operation. amount of calculation. Based on this, the present application provides a time-domain noise shaping method for audio signals.

The time-domain noise shaping process is described in detail below. In the following embodiments of this application, "and/or" describes the association relationship between associated objects, indicating that there can be three kinds of relationships, for example, A and/or B, it can mean that A exists alone, A and B exist at the same time, independently There is a case of B, where A, B can be singular or plural. The character "/" generally indicates that the associated objects are an "or" relationship. "At least one item(s) below" or similar expressions thereof refer to any combination of these items, including any combination of single item(s) or plural items(s). For example, at least one (a) of a, b or c may represent: a, b, c, a-b, a-c, b-c, or a-b-c, where a, b, c may be single or multiple. The singular expressions "a", "an", "the", "above", "the" and "the" are intended to also include such expressions as "one or more" unless the context clearly dictates otherwise. to the contrary. And, unless stated to the contrary, the ordinal numbers such as “first” and “second” mentioned in the embodiments of the present application are used to distinguish multiple objects, and are not used to limit the order, sequence, priority or priority of multiple objects. Importance. For example, the first task execution device and the second task execution device are only for distinguishing different task execution devices, and do not indicate the difference in priority or importance of the two task execution devices.

References in this specification to "one embodiment" or "some embodiments" and the like mean that a particular feature, structure or characteristic described in connection with the embodiment is included in one or more embodiments of the present application. Thus, appearances of the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in other embodiments," etc. in various places in this specification are not necessarily All refer to the same embodiment, but mean "one or more but not all embodiments" unless specifically emphasized otherwise. The terms "including", "including", "having" and their variants mean "including but not limited to" unless specifically emphasized otherwise.

As shown in FIG. 3 , the process of a time-domain noise shaping method for an audio signal provided by the present application, the method can be executed by an LC3 codec of an electronic device, and the electronic device can be a mobile phone and the above-mentioned FIG. 1 . Bluetooth headsets, etc., which are not specifically limited in this application. The electronic device may perform the following:

Step 301: Receive an audio signal, and convert the audio signal into a frequency domain signal.

For example, after receiving the audio signal, the above electronic device can sample, quantize, encode and convert the audio signal into a PCM signal, and then perform Fourier transform on the PCM signal, etc., to convert the PCM signal into a frequency domain signal.

Usually an electronic device will receive a segment of audio signal, which can be divided into multiple frames of audio signals according to a preset time (for example, audio data collected within 10 milliseconds (ms) is 1 frame), and each frame of audio signal is usually There is a one-to-one correspondence with the frequency domain signal, so how many frames of audio signals exist, that is, how many frames of frequency domain signals exist.

Step 302: Determine the k-th order prediction error value _ek of the frequency domain signal processed by the SNS; k is any integer from 1 to m; m is the filtering order of the filter in the time-domain noise shaping TNS.

Optionally, for the frequency-domain signal processed by the SNS, determine the k-th order normalized autocorrelation coefficient r _w (k) of the frequency-domain signal, and k is initialized to 1; based on the k-th order normalized autocorrelation coefficient r _w (k), determine the k-th order prediction error value e _k ; if the k-th order prediction error value e _k is not greater than the k-th order prediction error comparison value e_thresh _k , then assign the value of k plus 1 to k, and return to determine the frequency The step of normalizing the k-th order autocorrelation coefficient r _w (k) of the domain signal until the k-th order prediction error value e _k is greater than e_thresh _k or k is equal to m.

Optionally, the k-th order normalized autocorrelation coefficient r _w (k) can be calculated with reference to the following formula:

in,

Among them, X _s (n) is the input frequency domain signal, the superscript and subscript of the summation sign are the initial value and cutoff value of the filter, and f represents the number of filters (the number of filters used in different frequency bands is different), because the filter The filter will filter the three segments, so s is equal to 0-2. Based on the above formula 2, r ₀ (0) is 3.

Optionally, the electronic device performs calculation according to r _w (k), and the kth linear prediction coding value a(k); according to r _w (k) and a(k), determines the prediction error value _ek . In this application, the linear prediction coding value and the prediction error value may be determined by the following soft codes, but in practical application, other methods may also be used to determine, which is not specifically limited in this application. a(k) and e _k can be determined with reference to the following soft codes:

err=r ₀ (0)

a(0)=1

fork=1to8do

tmp(0)=1

forn=1tok-1do

tmp(n)=a(n)+rc(k)·a(k-n)

tmp(k)=rc(k)

forn=0tokdo

a(n)=tmp(n)

err(k)=(1-rc(k) ² ) err(k-1)

Among them, err is the prediction error value, rc is the intermediate value, and tmp() is the intermediate value.

Optionally, any order prediction error comparison value in each order prediction error comparison value is determined by the corresponding order prediction error in the sample data; the sample data is a frequency domain signal that does not need to be processed by TNS filtering.

For example, select a large number of audio signals (that is, sample data) that do not need to be processed by TNS, calculate the prediction error value of each order, and determine the largest prediction error value without TNS processing, then determine the prediction error value as the prediction of this order. The error comparison value, of course, in practical application, in order to improve the data calculation efficiency, the prediction error comparison value of each order can also be determined by fitting the sample data with a high-order polynomial. The prediction error comparison value determined in this way is more reliable, and the accuracy of data processing can be guaranteed.

The comparison value of each order prediction error is determined under the set loss accuracy condition. Among them, in the time-domain noise shaping, the loss of signal accuracy is different, and the corresponding prediction error comparison value is also different. Refer to Table 1 below to determine the prediction error comparison value. For example, when the processing order is ² and the signal loss accuracy is 20/00, the prediction error comparison value is _2.9992 . In practical applications, only one or more lines may be applied. , one or more columns, which is not limited in this application.

Table 1

Under different signal loss accuracy requirements, the granularity of calculation reduction in time-domain noise shaping processing is different. For example, when the signal accuracy is lossless, the calculation amount is reduced by 20%; when the signal accuracy _{loss is 10/00} ^, the calculation amount is reduced by 25%; When the signal accuracy loss is ^20/00 , the calculation amount is reduced by ³⁰ %; when the signal accuracy loss is _50/00 , the calculation amount is reduced by 35%, as shown in Figure 4, where the abscissa is the _yth linear prediction coding value, The ordinate is the prediction error comparison value. For example, when the signal accuracy is lossless, the comparison value of the prediction error corresponding to the 6th linear prediction coding value is 2.85. If the 6th prediction error value is less than 2.85, the temporal noise shaping operation is not performed. The necessary calculation reduces the amount of calculation, and further, the data processing efficiency can be improved.

In the present application, by setting the loop operation, the accuracy of data processing can be ensured, and the loop is terminated when e _k is greater than the k-th order prediction error comparison value e_threshk or k=m, which can reduce the amount of calculation and further improve the data processing efficiency.

In addition, for the frequency domain signal processed by SNS, the normalized autocorrelation coefficients of each order of the frequency domain signal can also be determined (that is, the normalized autocorrelation coefficients of the m order are directly calculated). order, the kth order prediction error value ek of the frequency domain signal is determined based on the normalized autocorrelation coefficients of order 1 to _k -1. In this way, there is no need to perform a loop operation, the prediction error values of each order can be directly determined, and the data processing efficiency is high.

Step 303: Determine the relationship between the k-th order prediction error value e _k and the k-th order prediction error comparison value e_thresh _k . If e _k is not greater than e_thresh _k , and the prediction error value of each order is not greater than the comparison value of prediction error of each order, go to step 304; if e _k is greater than e_thresh _k , go to step 305; Step 305 will be executed.

In addition, it should be noted that, in order to ensure the data processing accuracy, the judgment condition before executing step 304 may be that e _k is not greater than e_thresh _k , and the prediction error values of each order are not greater than the comparison value of prediction errors of each order.

Step 304 , filter the frequency domain signal through the filter in the TNS.

In step 305, the frequency domain signal is directly subjected to spectral quantization processing.

In this application, when the LC3 codec determines that the prediction error value of a certain order is greater than the prediction error comparison value, it does not perform filtering processing, and performs the operation of spectral quantization. In terms of the calculation of filtering processing, the calculation amount can be reduced, and the signal processing efficiency can obviously be improved when the calculation amount is reduced.

Before filtering the frequency-domain signal by the filter in the TNS, the prediction gain of the frequency-domain signal can also be determined based on the 0th-order normalized autocorrelation coefficient _rw (0) and the _mth -order prediction error value em; When it is determined that the prediction gain is greater than the first preset value, the frequency domain signal is filtered by the filter in the TNS.

The above-mentioned determination of the prediction gain _preGain (y) of the frequency domain signal according to r _w (0) and the prediction error value em can be determined by the following formula:

It should be noted that the first preset value may be 1.5, 2.5, etc., which is not specifically limited in this application. Of course, when the prediction gain is not greater than the first preset value, the frequency domain signal can be directly subjected to spectral quantization processing.

In this application, r _w (0) is determined after the frequency domain signal is normalized, and the predicted gain of the frequency domain signal is determined based on r _w (0); when the electronic device performs time domain noise processing, when the predicted gain is less than or equal to ( That is, when it is not greater than the first preset value, it is determined that the filtering operation is not performed, and the spectrum quantization operation is directly performed. Compared with the prior art, regardless of the relationship between the preset gain and the first preset value, the filtering operation must be performed. In terms of calculation, the amount of calculation can be reduced, and when the amount of calculation is reduced, the signal processing efficiency can obviously be improved.

Optionally, filter the frequency domain signal through a filter in the TNS, including:

Based on the m-order linear prediction coding value of the frequency domain signal, the m-1 order reflection coefficient is determined; the m-1 order reflection coefficient is quantized to obtain the m-1 order quantized reflection coefficient of the filter in the TNS; based on the m-1 order reflection coefficient The first-order quantized reflection coefficient is used to filter the frequency domain signal through the filter in the TNS.

当不满足上述条件时，可将γ设置为1，加权后的线性预测编码值为a_w(k)，可参照下述公式确定：In practical applications, if the predicted gain is less than or equal to the first preset value (1.5), the electronic device can set all the reflection coefficients to 0, and if the predicted gain is greater than the first preset value, it can also be performed based on the intermediate variable tns_lpc_weighting and the predicted gain. Judgment, and then determine the value of the reflection coefficient. For example, when it is determined that tns_lpc_weighting is 1, and the first preset value is greater than 1.5 and less than or equal to 2, set the weighting factor γ of the linear predictive coding value to be

When the above conditions are not met, γ can be set to 1, and the weighted linear prediction coding value is a _w (k), which can be determined with reference to the following formula:

a _w (k)=γ ^k a(k) fork=0....m

Formula 6

After the LC3 codec determines the weighted linear prediction coding value through the above-mentioned formula 6, the weighted linear prediction coding value a _w (k) can be converted into a reflection coefficient with reference to the following soft code:

tmp1(k)=a _w (k),k=0,...,m

fork=8to1do

rc(k-1)=tmp1(k)

e=(1-rc(k-1) ² )

forn=1tok-1do

forn=1tok-1do

tmp1(n)=tmp2(n)

Among them, err is the prediction error value, e is the intermediate value, rc() is the reflection coefficient, and tmp2() is the intermediate value.

After that, the LC3 codec quantizes the reflection coefficient by the following formula:

nint表示四舍五入到最接近的整数，rc_q(k,f)为量化后的反射系数，rc_i(k,f)为量化反射系数的中间值。in,

nint means rounding to the nearest integer, rc _q (k, f) is the quantized reflection coefficient, and rc _i (k, f) is the middle value of the quantized reflection coefficient.

It should also be noted that when quantizing the reflection coefficient, the LC3 codec can also calculate the bit overhead of TNS, so as to reduce the signal processing flow after TNS, where the bit overhead of TNS can be determined by the following formula:

with:

and:

Among them, ac_tns_order_bits is used to calculate the bit consumption of tns_order, that is, the position of the last non-zero value of the quantized reflection coefficient. ac_tns_coef_bits is used to calculate the bit consumption of quantized reflection coefficients.

In order to better illustrate the solution of the present application, it can be explained with reference to FIG. 5 . From the acquisition of the frequency domain signal for data processing, the LC3 codec can separately calculate m of the frequency domain signal (in FIG. 5, m is 8 as an example to Explain) normalized autocorrelation coefficients, and calculate 8 LPC coefficients a(k) based on the 8 normalized autocorrelation coefficients, and calculate the prediction error e _k based on the LPC coefficients and the normalized autocorrelation coefficients, and determine the prediction Whether the error e _k is less than or equal to the prediction error comparison value e_thresh _k , if not, perform spectrum quantization; if so, perform self-addition on k to obtain k+1, assign the value of k+1 to k, and determine k Whether +1 is greater than 8, if not, calculate the normalized autocorrelation function of the k+1th normalization process as r _w (k+1), and determine the frequency according to the 1st to kth normalized autocorrelation coefficients. The k+1th linear prediction coding value a(k+1) of the domain signal and the prediction error value e _k+1 , that is, the above steps are performed cyclically. If so, the LC3 codec calculates the prediction gain, determines the relationship between the prediction gain preGain and the first preset value (1.5), if it is not greater than the first preset value, performs spectrum quantization, if it is greater than the first preset value , then the reflection coefficient is calculated, and the reflection coefficient is quantized, the TNS bit consumption is calculated, and the quantized reflection coefficient is used to filter the frequency domain signal to obtain the filtered signal.

By setting the loop operation in the present application, the accuracy of data processing can be ensured, the calculation amount of the electronic device can be reduced, and the data processing efficiency can be further improved.

Based on the same concept, an embodiment of the present application provides a time-domain noise shaping device for audio signals, as shown in FIG.

Wherein, the error determination unit 601 is used to determine the k-th order prediction error value e _k of the frequency domain signal processed by the SNS; k is any integer from 1 to m; m is the filtering of the filter in the time domain noise shaping TNS order; the first error comparison unit 602 is used for performing spectrum quantization processing on the frequency domain signal if the k-th order prediction error value e _k is greater than the k-th order prediction error comparison value e_thresh _k ; the second error comparison unit 603 uses If the prediction error value of each order of the frequency domain signal is not greater than the corresponding prediction error comparison value of each order, the frequency domain signal is filtered by the filter in the TNS.

It should be noted that after receiving the audio signal in the present application, the electronic device can encode the audio signal to obtain a PCM signal, and then perform Fourier transform on the PCM signal to convert the PCM signal into a frequency domain signal for input to the LC3 Signal processing in the codec.

In this application, when the LC3 codec determines that the prediction error value of a certain order is greater than the prediction error comparison value, it does not perform filtering processing, and performs the operation of spectral quantization. In terms of the calculation of filtering processing, the calculation amount can be reduced, and the signal processing efficiency can obviously be improved when the calculation amount is reduced.

In an optional manner, the error determination unit 601 is specifically configured to, for the frequency domain signal processed by the SNS, determine the k-th order normalized autocorrelation coefficient r _w (k) of the frequency domain signal, where k is initialized to 1 ; Based on the k-th order normalized autocorrelation coefficient r _w (k), determine the k-th order prediction error value e _k ; if the k-th order prediction error value e _k is not greater than the k-th order prediction error comparison value e_thresh _k , then The value of k plus 1 is assigned to k, and the step of determining the k-th order normalized autocorrelation coefficient r _w (k) of the frequency domain signal is returned until the k-th order prediction error value e _k is greater than e_thresh _k or k is equal to m.

In the present application, by setting the loop operation, the accuracy of data processing can be guaranteed, and the loop can be ended when e _k is greater than the k-th order prediction error comparison value e_thresh _k or k=m, which can reduce the amount of calculation and further improve the data processing efficiency.

In an optional manner, the error determination unit 601 is specifically configured to determine the normalized autocorrelation coefficients of each order of the frequency domain signal for the frequency domain signal processed by the spectral noise shaping SNS; for any kth order, The kth order prediction error value _ek of the frequency domain signal is determined based on the 1st to k-1th order normalized autocorrelation coefficients.

In this way, the prediction error value of each order can be directly determined, and the data processing efficiency is high.

In an optional manner, before the second error comparison unit 603 performs filtering processing on the frequency domain signal through the filter in the TNS, the second error comparison unit 603 is further configured to use the zeroth-order normalized autocorrelation coefficient _rw (0) and the The _m -order prediction error value em determines the prediction gain of the frequency domain signal; when the prediction gain is greater than the first preset value, the frequency domain signal is filtered by the filter in the TNS.

According to the relationship between the predicted gain and the first preset value, the present application determines whether to start the filtering operation, and in this way, the accuracy of the data processing can be ensured.

In an optional manner, the second error comparison unit 603 is further configured to perform spectral quantization processing on the frequency domain signal when the prediction gain is not greater than the first preset value.

In this way, the calculation amount of time-domain noise processing can be reduced, and the data processing efficiency can be further improved.

In an optional manner, any order prediction error comparison value in each order prediction error comparison value is determined by the corresponding order prediction error in the sample data; the sample data is a frequency domain signal that does not need to be processed by TNS filtering.

The prediction error comparison value determined in this way is more reliable, and the accuracy of data processing can be guaranteed.

In an optional manner, the comparison values of prediction errors of each order are determined under a set loss accuracy condition.

In this application, the comparison values of prediction errors of each order are related to the loss accuracy, and this method can ensure the accuracy of data processing.

In an optional manner, the second error comparison unit 603 is specifically configured to, if each order prediction error value of the frequency domain signal is not greater than the corresponding each order prediction error comparison value, compare the frequency domain The signal is filtered.

In this way, the accuracy of data processing can be guaranteed.

In an optional manner, the second error comparison unit 603 is specifically configured to determine the m-1-order reflection coefficient based on the m-order linear prediction coding value of the frequency domain signal; perform quantization processing on the m-1-order reflection coefficient to obtain The m-1 order quantized reflection coefficient of the filter in the TNS; based on the m-1 order quantized reflection coefficient, the frequency domain signal is filtered by the filter in the TNS.

After the temporal noise shaping apparatus for audio signals in the exemplary embodiment of the present application is introduced, next, the computing device of another exemplary embodiment of the present application is introduced.

As will be appreciated by one skilled in the art, various aspects of the present application may be implemented as a system, method or program product. Therefore, various aspects of the present application can be embodied in the following forms, namely: a complete hardware implementation, a complete software implementation (including firmware, microcode, etc.), or a combination of hardware and software aspects, which may be collectively referred to herein as implementations "circuit", "module" or "system".

In some possible implementations, a computing device according to the present application may include at least one processor, and at least one memory. The memory stores a computer program that, when executed by the processor, causes the processor to execute the steps in the temporal noise shaping method for audio signals according to various exemplary embodiments of the present application described above in this specification. For example, the processor may perform steps 301-305 as shown in FIG. 3 .

A computing device 130 according to such an embodiment of the present application is described below with reference to FIG. 7 . The computing device 130 shown in FIG. 7 is only an example, and should not impose any limitations on the functions and scope of use of the embodiments of the present application. As shown in FIG. 7 , the computing device 130 takes the form of a general smart terminal (or a Bluetooth headset). Components of the computing device 130 may include, but are not limited to, the above-mentioned at least one processor 131 , the above-mentioned at least one memory 132 , and a bus 133 connecting different system components (including the memory 132 and the processor 131 ).

Bus 133 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, a processor, or a local bus using any of a variety of bus structures. Memory 132 may include readable media in the form of volatile memory, such as random access memory (RAM) 1321 and/or cache memory 1322 , and may further include read only memory (ROM) 1323 . The memory 132 may also include a program/utility 1325 having a set (at least one) of program modules 1324 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, which An implementation of a network environment may be included in each or some combination of the examples.

Computing device 130 may also communicate with one or more external devices 134 (eg, keyboards, pointing devices, etc.), and/or with any device that enables the computing device 130 to communicate with one or more other smart terminals (eg, routers, modem, etc.) communication. Such communication may take place through input/output (I/O) interface 135 . Also, the computing device 130 may communicate with one or more networks (eg, a local area network (LAN), a wide area network (WAN), and/or a public network such as the Internet) through a network adapter 136 . As shown, network adapter 136 communicates with other modules for computing device 130 via bus 133 . It should be understood that, although not shown, other hardware and/or software modules may be used in conjunction with computing device 130, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID systems, tape drives and data backup storage systems.

In some possible implementations, various aspects of the temporal noise processing method provided by the present application can also be implemented in the form of a program product, which includes a computer program, and when the program product runs on a computer device, the computer program is used for The computer device is caused to perform the steps in the temporal noise shaping method for an audio signal according to various exemplary embodiments of the present application described above in this specification. For example, the processor may perform steps 301-305 as shown in FIG. 3 .

The program product may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus or device, or a combination of any of the above. More specific examples (non-exhaustive list) of readable storage media include: electrical connections with one or more wires, portable disks, hard disks, random access memory (RAM), read only memory (ROM), erasable programmable read only memory (EPROM or flash memory), optical fiber, portable compact disk read only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

The program product for temporal noise processing of the embodiments of the present application may adopt a portable compact disk read only memory (CD-ROM) and include a computer program, and may run on a smart terminal. However, the program product of the present application is not limited thereto, and in this document, the readable storage medium may be any tangible medium containing or storing a program that can be used by or combined with an instruction execution system, apparatus or device.

A readable signal medium may include a propagated data signal in baseband or as part of a carrier wave with a readable computer program embodied thereon. Such propagated data signals may take a variety of forms including, but not limited to, electromagnetic signals, optical signals, or any suitable combination of the foregoing. A readable signal medium can also be any readable medium, other than a readable storage medium, that can transmit, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.

It should be noted that although several units or sub-units of the apparatus are mentioned in the above detailed description, this division is merely exemplary and not mandatory. Indeed, according to embodiments of the present application, the features and functions of two or more units described above may be embodied in one unit. Conversely, the features and functions of one unit described above may be further subdivided to be embodied by multiple units.

Furthermore, although the operations of the methods of the present application are depicted in the figures in a particular order, this does not require or imply that the operations must be performed in the particular order, or that all illustrated operations must be performed to achieve desirable results. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined to be performed as one step, and/or one step may be decomposed into multiple steps to be performed.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present application. It will be understood that each flow and/or block in the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to the processor of a general purpose computer, special purpose computer, embedded processor or other programmable access frequency prediction device to produce a machine for execution by the processor of the computer or other programmable access frequency prediction device The instructions produce means for implementing the functions specified in a flow or flow of the flowchart and/or a block or blocks of the block diagram.

These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable access frequency prediction device to operate in a particular manner, such that the instructions stored in the computer readable memory result in an article of manufacture comprising the instruction means, The instruction means implements the functions specified in the flow or flows of the flowcharts and/or the block or blocks of the block diagrams.

These computer program instructions may also be loaded on a computer or other programmable access frequency prediction device, such that a series of operational steps are performed on the computer or other programmable device to produce a computer-implemented process, thereby executing on the computer or other programmable device Execution of the instructions provides steps for implementing the functions specified in the flowchart or blocks and/or the block or blocks of the block diagrams.

While the preferred embodiments of the present application have been described, additional changes and modifications to these embodiments may occur to those skilled in the art once the basic inventive concepts are known. Therefore, the appended claims are intended to be construed to include the preferred embodiment and all changes and modifications that fall within the scope of this application.

Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims (12)

1. A time-domain noise shaping method for an audio signal, wherein the method comprises: Determine the k-th order prediction error value e _k of the frequency-domain signal processed by the frequency-domain noise shaping SNS; k is any integer from 1 to m; the m is the filtering order of the filter in the time-domain noise shaping TNS; If the k-th order prediction error value _ek is greater than the k-th order prediction error comparison value e_thresh _k , performing spectral quantization processing on the frequency domain signal; If the prediction error value of each order of the frequency domain signal is not greater than the corresponding prediction error comparison value of each order, the frequency domain signal is filtered by the filter in the TNS. 2. The method according to claim 1, wherein the determining the kth order prediction error value _ek of the frequency domain signal processed by the frequency domain noise shaping SNS comprises: For the frequency domain signal processed by SNS, determine the k-th order normalized autocorrelation coefficient r _w (k) of the frequency domain signal, and the k is initialized to 1; Based on the k-th order normalized autocorrelation coefficient r _w (k), determine the k-th order prediction error value e _k ; If the k-th order prediction error value e _k is not greater than the k-th order prediction error comparison value e_thresh _k , assign the value of k plus 1 to the k, and return the determination of the frequency domain signal. A step of normalizing the autocorrelation coefficient r _w (k) of order k, until the k-th order prediction error value _ek is greater than the e_thresh _k or the k is equal to the m. 3. The method according to claim 1, wherein the determining the kth order prediction error value _ek of the frequency domain signal processed by the frequency domain noise shaping SNS comprises: For the frequency domain signal processed by the SNS, determine the normalized autocorrelation coefficients of each order of the frequency domain signal; For any kth order, the kth order prediction error value _ek of the frequency domain signal is determined based on the 1st to k-1th order normalized autocorrelation coefficients. 4. The method according to claim 1, characterized in that, before performing filtering processing on the frequency domain signal by a filter in the TNS, the method further comprises: determining the prediction gain of the frequency domain signal based on the 0th order normalized autocorrelation coefficient r _w (0) and the _mth order prediction error value em; When the prediction gain is greater than a first preset value, the frequency domain signal is filtered by a filter in the TNS. 5. The method according to claim 4, wherein the method further comprises: When the prediction gain is not greater than the first preset value, spectral quantization processing is performed on the frequency domain signal. 6. The method according to any one of claims 1-5, wherein any order prediction error comparison value in the various order prediction error comparison values is determined by corresponding order prediction errors in the sample data, so The sample data is a frequency domain signal that does not need to be processed by TNS filtering. 7. The method according to any one of claims 1-5, wherein the comparison value of each order prediction error is determined under a set loss accuracy condition. 8. The method according to any one of claims 1-5, wherein, if the prediction error value of each order of the frequency domain signal is not greater than the corresponding comparison value of each order prediction error, then pass the TNS The filter in performs filtering processing on the frequency domain signal, including: If each order prediction error value of the frequency domain signal is not greater than the corresponding comparison value of each order prediction error, the frequency domain signal is filtered by the filter in the TNS. 9. The method according to any one of claims 1-5, wherein the filtering the frequency domain signal by a filter in the TNS comprises: determining the m-1 order reflection coefficient based on the m-order linear prediction coding value of the frequency domain signal; Quantizing the m-1 order reflection coefficient to obtain the m-1 order quantized reflection coefficient of the filter in the TNS; Based on the m-1 order quantized reflection coefficient, the frequency domain signal is filtered by a filter in the TNS. 10. A time-domain noise shaping device for audio signals, wherein the device comprises: An error determination unit, configured to determine the k-th order prediction error value e _k of the frequency domain signal processed by the frequency domain noise shaping SNS; k is any integer from 1 to m; the m is the filter in the time domain noise shaping TNS filter order of the filter; a first error comparison unit, configured to perform spectral quantization processing on the frequency domain signal if the k-th order prediction error value e _k is greater than the k-th order prediction error comparison value e_thresh _k ; The second error comparison unit is configured to filter the frequency domain signal through the filter in the TNS if the prediction error value of each order of the frequency domain signal is not greater than the corresponding prediction error comparison value of each order. 11. A computing device, comprising: a memory and a processor; memory for storing program instructions; The processor is configured to call the program instructions stored in the memory, and execute the method according to any one of claims 1-9 according to the obtained program. 12. A computer storage medium storing computer-executable instructions, wherein the computer-executable instructions are used to perform the method according to any one of claims 1-9.

Images