WO2014005473A1 - 一种送受话端采样率偏差纠正方法和系统 - Google Patents
一种送受话端采样率偏差纠正方法和系统 Download PDFInfo
- Publication number
- WO2014005473A1 WO2014005473A1 PCT/CN2013/076487 CN2013076487W WO2014005473A1 WO 2014005473 A1 WO2014005473 A1 WO 2014005473A1 CN 2013076487 W CN2013076487 W CN 2013076487W WO 2014005473 A1 WO2014005473 A1 WO 2014005473A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- sampling
- sending
- sampling rate
- time
- receiving end
- Prior art date
Links
- 238000005070 sampling Methods 0.000 title claims abstract description 323
- 238000000034 method Methods 0.000 title claims abstract description 40
- 230000005540 biological transmission Effects 0.000 claims abstract description 82
- 238000012546 transfer Methods 0.000 claims abstract description 47
- 238000012937 correction Methods 0.000 claims abstract description 23
- 238000012952 Resampling Methods 0.000 claims 2
- 230000001934 delay Effects 0.000 claims 2
- 238000004364 calculation method Methods 0.000 abstract description 7
- 238000010586 diagram Methods 0.000 description 6
- 230000000694 effects Effects 0.000 description 6
- 238000004891 communication Methods 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 230000001629 suppression Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000003139 buffering effect Effects 0.000 description 2
- 238000012886 linear function Methods 0.000 description 2
- 238000007476 Maximum Likelihood Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000003780 insertion Methods 0.000 description 1
- 230000037431 insertion Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04M—TELEPHONIC COMMUNICATION
- H04M9/00—Arrangements for interconnection not involving centralised switching
- H04M9/08—Two-way loud-speaking telephone systems with means for conditioning the signal, e.g. for suppressing echoes for one or both directions of traffic
- H04M9/082—Two-way loud-speaking telephone systems with means for conditioning the signal, e.g. for suppressing echoes for one or both directions of traffic using echo cancellers
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
- G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
Definitions
- the present invention relates to the field of audio processing technologies, and in particular, to a method and system for correcting a sample rate deviation of a receiving and receiving end.
- BACKGROUND OF THE INVENTION In voice communication, in order to ensure call quality and device security, echo cancellation is usually performed in voice communication.
- the echo path filter and the echo signal are calculated by the two, and the echo signal is eliminated from the signal of the sending end to avoid echo interference. communication.
- the existing scheme is to calculate the sampling rate deviation of the two ends in the sampling clock period of the transmitting and receiving end signals for a period of time, or the variation range of the pre-set sampling rate deviation. Within the default, the sampling rate deviation is within 20 Hz, and the sampling rate deviation is calculated by the pure algorithm. After calculating the sample rate deviation, the existing scheme usually uses the calculated sample rate difference to the echo cancellation filter, and the echo cancellation system adjusts accordingly to correct the sample rate deviation and ensure the echo cancellation effect. .
- the existing sampling rate deviation correction scheme has at least the following defects:
- the existing solution In calculating the sampling rate deviation, the existing solution needs to monitor the sending and receiving end sample clock. In some cases, special hardware settings are required, such as setting a high-performance CPU, which requires high hardware, and is computationally cumbersome. It also occupies more storage resources. Moreover, the existing solution does not directly correct the sample rate deviation, but only transmits the sample rate deviation to the echo cancellation system.
- the echo cancellation system performs adjustment processing, and the echo cancellation system adjusts the manner to increase echo cancellation. The burden of the system affects the effect of echo cancellation.
- the present invention provides a method and system for correcting the sample rate deviation of a transmitting and receiving terminal, so as to solve the problem that the existing solution has high requirements on hardware setting, or is complicated in calculation and narrow in application scope, and is not directly in the existing solution.
- the problem of the echo cancellation system being burdened by the correction of the sample rate deviation is relatively heavy.
- the embodiment of the invention provides a method for correcting the sample rate deviation of the sending and receiving end, the method comprising: calculating a transfer function of the signal of the receiving end with respect to the signal of the sending end according to the signal of the sending and receiving end; The transfer function obtains the transmission delay of the receiving end at each sampling time; using the linear relationship between the transmission delay and the transmission delay and the sampling rate deviation, the parameter fitting method is used to obtain the receiving time of each sampling time.
- Deviation of sampling rate adjusting the sampling rate of the signal of the sending end or the receiving end of each sampling time according to the deviation of the sampling rate, realizing the sampling rate correction, and correcting the sending end signal with the same sampling rate
- the receiver signal is input into the echo cancellation system, and is used for the echo cancellation system to directly perform echo cancellation using the same terminal signal and the receiver signal with the corrected sample rate.
- the embodiment of the invention further provides a sending and receiving end sample rate deviation correction system, the system comprising: a delay estimator, a sample rate deviation estimator and a sample rate adjuster, the input of the delay estimator And the output end of the delay estimator is connected to the input end of the sample rate deviation estimator, and the output end of the sample rate deviation estimator is connected to the ⁇ An input end of the sample rate adjuster, the input end of the sample rate adjuster is further connected to the sending end signal or the receiving end signal, and the output end of the sample rate adjuster is connected to the echo canceling system,
- the delay estimator is configured to calculate a transfer function of the received end signal with respect to the sending end signal according to the sending and receiving end signal; and use the transfer function to obtain the transmission delay of the sending and receiving end of each sample time ;
- the sampling rate deviation estimator is configured to use the linear relationship between the transmission delay and the transmission delay and the sampling rate deviation, and use the parameter fitting method to obtain the sampling rate deviation of the sending and receiving ends at each sampling time
- the sampling rate adjuster is configured to adjust the sampling rate of the sending end signal or the receiving end signal according to the sampling rate deviation, to achieve the sampling rate correction, and correct the sampling rate
- the same transmitting end signal and the receiving end signal are input into the echo canceling system for the echo canceling system to directly perform echo cancellation using the same transmitting end signal and the receiving end signal with the corrected sampling rate.
- the embodiment of the invention utilizes the characteristic that the transmission delay and the sampling rate deviation have a linear relationship, and the transmission delay between the transmitting and receiving ends is obtained based on the signal of the transmitting and receiving end, and the parameters are matched between the sending and receiving ends.
- the technical means of sampling rate deviation can obtain high-precision sampling rate deviation in real time, without additional hardware overhead, simple calculation method and low system cost. Further, since the scheme uses the technical means for correcting the sample rate deviation before the echo canceling operation, the burden of the echo canceling system is reduced, and the quality of the echo canceling is improved.
- FIG. 1 is a flowchart of a method for correcting a sampling rate deviation of a transmitting and receiving terminal according to a first embodiment of the present invention
- FIG. 2A is a transmission delay and a sampling time when a sample rate deviation is constant according to an embodiment of the present invention
- 2B is a schematic diagram showing a relationship between a transmission delay and a sampling time when a sample rate deviation is changed according to an embodiment of the present invention
- FIG. 3 is a schematic structural diagram of a sample rate deviation correction system according to Embodiment 2 of the present invention
- FIG. 4 is a schematic structural diagram of another sample rate deviation correction system according to Embodiment 2 of the present invention.
- the experimental results of the echo cancellation effect before and after the correction of the sampling rate provided by the embodiment are shown.
- DETAILED DESCRIPTION OF THE EMBODIMENTS In order to make the objects, technical solutions, and advantages of the present invention more comprehensible, the embodiments of the present invention will be further described in detail below.
- a method for correcting a sampling rate deviation of a receiving and receiving terminal includes:
- the embodiment of the present invention utilizes the characteristic that there is a linear relationship between the transmission delay and the sampling rate deviation, and the following principle is used to illustrate the principle that the sampling rate deviation can be corrected by using the above features:
- the relative transmission delay between the receiving end and the transmitting end is a linear function of the sampling time.
- the linear function can be expressed as follows:
- the frequency of the receiving end is the frequency of the sending end, and ⁇ 3 ⁇ 4 is the deviation of the sampling rate.
- the transmission delay of the receiving end at time n is expressed as ) W , then ) W and n meet the following linear relationship:
- C is a constant, determined by the transmission environment.
- Fig. 2A a schematic diagram showing the relationship between the sample and the n when the sample rate deviation is constant, the abscissa is the sample time, the ordinate is the transmission delay, and the slope of the line in the figure is the sample rate deviation. When the sample rate deviation changes, the slope also changes.
- Fig. 2B is a diagram showing the relationship between [ «] and n when the sample rate deviation changes (non-constant).
- dFs can be estimated from [[] and n, and the signal of the terminal or the signal of the terminal is corrected according to ⁇ 3 ⁇ 4. If the estimation is real-time, when the sampling rate deviation changes, the change can be tracked and adapted, so that the sampling rate deviation of the transmitting and receiving terminals can be corrected online in real time.
- the embodiment of the invention utilizes the characteristic that the transmission delay and the sampling rate deviation have a linear relationship, and the transmission delay between the transmitting and receiving ends is obtained based on the signal of the transmitting and receiving end, and the parameters are matched between the sending and receiving ends.
- the technical means of sampling rate deviation can obtain high-precision sampling rate deviation in real time, without additional hardware overhead, simple calculation method and low system cost. Further, since the solution uses the technical means for correcting the sample rate deviation before the echo cancellation operation, the burden of the echo cancellation system is reduced. Improve the quality of echo cancellation.
- the foregoing step 11 in the embodiment specifically includes: using each of the current sampling times of each sampling time, using the current sending end signal and the current sampling time The predetermined number of the sending end signals before the sampling time generates the sending end data frame of the current sampling time; generating the current data by using the receiving end signal of the current sampling time and a predetermined number of the receiving end signals before the current sampling time The receiving end data frame of the moment; calculating the transfer function of the receiving end signal with respect to the sending end signal at the current sampling time by using the sending end data frame and the receiving end data frame of the current sampling time.
- step 11 the transfer function of the received end signal of the current sampling time relative to the sending end signal is calculated by using the sending end data frame and the receiving end data frame of the current sampling time by the following formula: :
- h is the transfer function
- ⁇ ] is the frequency domain form of the data frame of the receiving end of the current sampling time n
- ⁇ ] is the frequency domain form of the data frame of the transmitting end of the current sampling time n
- H is The frequency domain form of the transfer function h is a conjugate of ⁇ ]
- E (.) represents the desired operation
- t(.) represents the inverse Fourier transform.
- the foregoing step 12 specifically includes: selecting, for each current sampling time of each sampling time, a time point corresponding to a maximum value of the absolute value of the transfer function of the current sampling time, as the current sampling time is sent to the receiving end.
- the transmission delay estimation value is obtained according to the transmission delay estimation value, for example, the obtained transmission delay estimation value is directly used as the transmission delay of the receiving end at the current sampling time used.
- the foregoing step 13 specifically includes: generating, by using the current transmission time delay of the current sampling time and the transmission delay of the predetermined number of transmitting and receiving ends before the current sampling time, for each current sampling time of each sampling time
- the foregoing step 14 specifically includes: re-sampling the signal of the sending end according to the sampling rate deviation of the sending end of the current sampling time for each current sampling time of each sampling time, At the current sampling time, the sending end signal is the same as the sampling rate of the receiving end signal;
- the receiving end signal is re-sampled by the interpolation method, and the same sampling rate as that of the sending end signal is obtained at the current sampling time.
- the above interpolation method may be a polynomial interpolation method or a linear interpolation method or the like.
- the second embodiment of the present invention together with the specific implementation device, describes the correcting scheme of the sample rate deviation of the sending and receiving end provided by the embodiment.
- a sample rate deviation correction system which includes a delay estimator 1, a sample rate deviation estimator 2, and a sample rate adjuster 3, which is subjected to Both the voice signal and the voice signal are digitally transmitted, so correspondingly, an analog-to-digital converter is also provided at the transmitting end to convert the analog signal y ( t ) of the collected voice terminal into a digital signal.
- the transmitter (digital) signal y[n] and the receiver (digital) signal x[n] are connected to the input of the delay estimator 1, and the output of the delay estimator 1 is connected to the sample rate deviation estimator At the input of 2, the transmission delay [["] output from the output of the delay estimator 1 is transmitted to the input of the sample rate deviation estimator 2.
- the output of the sample rate deviation estimator 2 is connected to the input of the sample rate adjuster 3, and the sample rate deviation output from the output of the sample rate deviation estimator 2 is transmitted to the sample rate adjuster 3. At the input end, the output of the sample rate adjuster 3 is connected to the echo cancellation system.
- the sampling rate deviation obtained in the scene shown in FIG. 3 corrects the sending end signal to obtain the same receiving and receiving end signal, so the input of the sample rate adjuster in the scene.
- the terminal also accesses the sending end signal, and the corrected transmitting end signal and the receiving end signal x[n] are transmitted to the input end of the echo cancellation system.
- the sampling frequency of the transmitting end signal is corrected by the sampling rate adjuster.
- the sampling rate adjuster can also be used to detect the signal of the receiving end signal.
- the sample frequency is corrected, and the case of the main former in this embodiment is taken as an example for explanation.
- the input end of the sample rate adjuster does not need to access the signal of the sending end, but needs to access the signal of the receiving end, and the message will be sent in this scenario.
- the end signal y[n] and the corrected received end signal c'[ «] ] are transmitted to the input of the echo cancellation system.
- the delay estimator 1 is configured to calculate a transfer function of the received end signal with respect to the sending end signal according to the sending and receiving end signal; and use the transfer function to obtain the transmission delay of the sending and receiving end of each sample time ;
- the sampling rate deviation estimator 2 is configured to use the linear relationship between the transmission delay and the transmission delay and the sampling rate deviation, and use the parameter fitting method to obtain the sampling rate deviation of the sending and receiving ends at each sampling time.
- the sampling rate adjuster 3 is configured to adjust the sampling rate of the sending end signal or the receiving end signal according to the sampling rate deviation, thereby realizing the sample rate deviation correction, and correcting the sample
- the same terminal signal and the receiver signal are input to the echo cancellation system for echo cancellation. It can be seen from the above that the sampling rate of the sending end signal or the receiving end signal is corrected and then sent to the echo canceling system, that is, the input quantity of the echo canceling system is the sending and receiving end signal having the same sampling frequency. For example, the method of obtaining the echo signal with the same sampling frequency and then performing the echo cancellation helps to improve the echo cancellation effect and reduces the burden of the echo cancellation system located at the back end of the data processing.
- the operation of this part is mainly implemented by the delay estimator.
- the digital signal y[n] of the transmitting end and the digital signal x[n] of the receiving end pass through the buffer to form the data frame and the receiving end of the transmitting end respectively.
- the data frame ⁇ [ ⁇ ] is expressed as follows:
- the delay estimator 1 includes a receiver side buffer 11, a source terminal buffer 12, a transfer function estimator 13 and a delay calculator 14.
- the sending end buffer 12 is configured to buffer the sending end signal of each sample moment.
- the receiving end buffer 11 is configured to buffer the received end signal of each sample time;
- the transfer function estimator 13 is configured to use the sending end signal of the current sampling time in the sending end buffer and the predetermined number of sending ends before the current sampling time for each current sampling time of each sampling time.
- the signal generates a data frame of the sending end at the current sampling time; using the current sampling time in the receiving end buffer Receiving end signal and a predetermined number of receiving end signals before the current sampling time to generate a receiving end data frame of the current sampling time; and using the sending end data frame and the receiving end data frame by using the current sampling time Calculating a transfer function of the received end signal at the current sampling time relative to the signal of the transmitting end;
- the predetermined number of the above is L-1, and the transfer function estimator 13 of the immediate delay estimator 1 generates the current using the send end signal y[n] of the current sample time and the L-1 call end signals before the current sample time.
- the data frame of the sending end of the sample time and the received end data frame of the current sampling time are generated by using the receiving end signal x[n] of the current sampling time and the L-1 receiving end signals before the current sampling time.
- the specific value of L is related to the output delay limit of the system. For example, L can take 256 or 512.
- the data frame is sent to the input of the transfer function estimator 13 to calculate the transfer function h of «].
- the transfer function estimator 13 calculates the transfer function of the call end signal of the current sample time with respect to the signal of the sending end by using the data frame of the sending end and the data frame of the receiving end of the current sampling time.
- the transfer function calculation method can use the cross-power spectrum and the self-power word division method. The specific formula is as follows:
- h is the transfer function
- ⁇ ] is the frequency domain form of the data frame j [ «] of the receiving end at the current sampling time n
- ⁇ ] is the frequency domain form of the data frame of the transmitting terminal at the current sampling time n
- H is the frequency domain form of the transfer function h, which is the conjugate of ⁇ ]
- E (.) represents the expectation operation
- t(.) represents the inverse Fourier transform.
- the transfer function input delay calculator 14 calculated by the transfer function estimator 13 selects the maximum value of the absolute value of the transfer function at the current sample time for each current sample time of each sample time.
- the corresponding time point is used as the transmission delay estimation value of the current receiving time at the receiving end; and, according to the transmission delay estimation value, the transmission delay of the current receiving time and the receiving end used is obtained.
- the value of the h parameter (time parameter) corresponding to the maximum value of the absolute value of h is selected as the estimated value of the signal delay of the transmitting and receiving ends. From above, by the calculation of the delay estimator 1, the corresponding transmission and reception end transmission delay D £ W can be calculated for each sampling time n.
- the value of the signal before the initial sampling time can be ⁇ Use the default value (such as 0).
- the sample rate deviation estimator includes a delay buffer 21 and a delay sample rate deviation fitter 22.
- the calculated transmission delay [ «] is sent to the delay buffer 21 of the sample rate deviation estimator, and the data frame formed by the sample time corresponding to the transmission delay data frame is recorded as:
- n ⁇ n - M + ⁇ ⁇ n - ⁇ n)
- M is the length of the data frame.
- M embodies the length of the observation time.
- the increase of the observation time can improve the accuracy of the fitting.
- the delay buffer 21 in the sample rate deviation estimator 2 is used to buffer the transmission delay of the receiving end of each sample time.
- the delay sample rate deviation fitter 22 is configured to use the current transmission time delay of the current sampling time in the delay buffer and the predetermined number before the current sampling time for each current sampling time of each sampling time.
- the transmission delay of the transmitting and receiving ends generates a transmission delay data frame of the current sampling time; and, according to a linear relationship between the transmission delay and the sampling rate deviation, each element in the transmission delay data frame is compared with each The parameter fitting is performed at the time of sampling, and the sampling rate deviation of the current receiving end is obtained.
- the instantaneous delay rate deviation fitter 22 generates the current sampling time by using the transmission delay D E [n] sent to the receiving end at the current sampling time n and the transmission delay of the M-1 sending and receiving ends before the current sampling time. Transmitting the time delay data frame, then using the delay sample rate deviation fitter 22 to compare the elements in the transmission delay data frame with respect to each sample time according to a linear relationship between the transmission delay and the sample rate deviation The parameter fitting is performed to obtain the sampling rate deviation of the current receiving time at the receiving end. The delay and the delay ratio of the sample are fed to the fitter, and the slope of the pair is estimated. This slope is the estimated sample rate deviation dFs E .
- the fitting method can be a least squares fitting, a maximum likelihood fitting or other parameter fitting method.
- the specific formula can be expressed as follows: ⁇ ⁇ k - ⁇ ⁇ ) (D E [k] - E (D E [n]))
- the sampling rate deviation can be calculated in real time for each sampling time n.
- the operation of this part is mainly realized by the sample rate adjuster 3, and the sample rate adjuster 3 corrects the sampling rate of the signal of the sending end or the signal of the receiving end according to the sampling rate deviation, so as to eliminate the echo.
- the system uses the sent-end signal and the received-end signal with the same sample rate after correction to perform echo cancellation.
- the sampling rate adjuster 3 re-samples the sending end signal or the receiving end signal of the current sampling time according to the sampling rate deviation of the current sampling time, and obtains the current sample.
- the sending end signal and the receiving end signal have the same sampling rate.
- the sampling rate adjuster 3 includes a re-sampling buffer 31.
- the sampling rate adjuster 3 includes a re-sampling buffer 31 for buffering the signal of the transmitting end at each sampling time.
- the sampling rate adjuster 3 is specifically configured for each of the sampling times.
- the interpolation signal is re-sampled in the interpolation mode to obtain the sampling rate at the current sampling time.
- the sending end signal is the same as the receiving end signal; when the input end of the sampling rate adjuster 3 is connected to the receiving end signal, the sampling rate adjuster 3 includes the receiving end signal for buffering each sample time At this time, the sample rate adjuster 3 is specifically used for each current sample time of each sample time, according to the sampling rate deviation of the current receiving time at the time of the sample, The insertion mode re-receives the received signal in the re-sampling buffer 31 Bian-like line, terminal signals to obtain samples of the current Bian Bian comp same time preclude the transmitting terminal signal sample rate.
- the embodiment is separately set for the delay estimator 1
- L. represents a rounding down operation, indicating the signal of the sending end of the sample time n after the sample is repeated.
- the signal of the receiving end after repeated sampling can be expressed as follows: l + dFs E ) + l - [
- L. represents the rounding down operation
- x'W represents the received end signal of the sample time n after the repeated sample.
- the device in this embodiment does not require additional special settings.
- the above-described delay estimator 1 can be implemented by a buffer, a transfer function estimator 13 and a delay calculator 14, and the transfer function estimator 13 can be composed of a multiplier, Implemented by an integrator, divider, and inverse Fourier transformer.
- the delay calculator 14 can be implemented by a multiplier, a divider, and a comparator, and the like.
- the present embodiment corrects the sampling rate deviation based on the signal of the transmitting end, and does not need to monitor the clock of the transmitting end, thereby eliminating the need for additional hardware settings, reducing the hardware performance requirements and saving system cost.
- the embodiment of the present invention utilizes the characteristic that there is a linear relationship between the transmission delay and the sampling rate deviation, and uses the signal based on the sending and receiving end to obtain the transmission delay between the transmitting and receiving ends, and the parameters are fitted to the receiving and receiving end.
- the technical means of deviation between sample rates can obtain high-precision sample rate deviation in real time without additional hardware overhead, simple calculation method and low system cost.
- the scheme uses the technical means for correcting the sample rate deviation before the echo canceling operation, the burden of the echo canceling system is reduced, and the quality of echo cancellation is improved.
- Fig. 5 is a graph showing experimental results of the echo canceling effect before and after the correction of the sampling rate.
- the expected sampling rate of the receiving end is 16000 Hz, and the deviation of the sampling rate is 0.9 Hz.
- the sampling rate deviation correction is performed from 29 seconds.
- the transmitting and receiving end signal having the sampling rate deviation is input to the echo canceling system before 29 seconds, and the transmitting and receiving end signals having the same sampling rate obtained by the present scheme are input to the echo canceling system after 29 seconds.
- the ordinate is RMS (mean square error) Power (power)
- the abscissa is time
- the dash in the figure is the energy curve of the Mic signal
- the solid line is the output of the echo cancellation system.
- the residual echo energy curve, the difference between the two curves is the echo suppression amount.
Landscapes
- Engineering & Computer Science (AREA)
- Signal Processing (AREA)
- Computational Linguistics (AREA)
- Quality & Reliability (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Cable Transmission Systems, Equalization Of Radio And Reduction Of Echo (AREA)
- Telephone Function (AREA)
Abstract
本发明公开了一种送受话端采样率偏差纠正方法和系统，能够实时得到高精度的采样率偏差，并对送受话端信号进行采样率纠正，得到纠正后的采样率相同的送话端信号和受话端信号送入回声消除系统进行回声消除。本发明有助于提高回声消除的质量，且计算方法简单，成本较低。本发明实施例提供的一种送受话端采样率偏差纠正方法包括：根据送受话端信号计算各采样时刻的受话端信号相对于送话端信号的传递函数；利用传递函数获取各采样时刻送受话端的传输时延；利用传输时延和传输时延与采样率偏差之间的线性关系，采用参数拟合方式得到各采样时刻送受话端的采样率偏差；根据采样率偏差调整各采样时刻送话端信号或受话端信号的采样率。
Description
一种送受话端采样率偏差纠正方法和系统
技术领域 本发明涉及音频处理技术领域， 特别涉及一种送受话端釆样率偏差纠正方 法和系统。 背景技术 语音通讯中， 为保证通话质量和设备安全， 通常会在语音通讯中做回声消 除。 目前常用的回声消除方法中， 当受话端信号、 送话端信号已知时， 通过二 者计算出回声路径滤波器以及回声信号， 并将回声信号从送话端信号中消去， 避免回声干扰通讯。
然而， 当今大多数通讯已经实现数字化， 受话端信号和送话端信号都是以 数字方式传输的。 由于釆样时钟的不同， 送受话端信号可能存在釆样率差异， 釆样率差异会降低回声路径滤波器以及回声信号的估计精度， 导致回声消除性 能下降。
为了降低或消除釆样率偏差对回声消除性能的影响， 需要在回声消除之前 计算出送受话端信号之间的釆样率偏差并进行纠正。 在计算釆样率偏差时， 现 有方案的做法是， 统计一段时间内送受话端信号的釆样时钟周期计算得到两端 的釆样率偏差， 或者是在预先设置的釆样率偏差的变动范围内， 默认釆样率偏 差在 20Hz以内， 釆用纯算法的方式， 计算出釆样率偏差。 在计算出釆样率偏差 后， 现有方案通常釆用将计算得到的釆样率差异传递给回声消除滤波器， 由回 声消除系统进行相应的调节， 以纠正釆样率偏差， 保证回声消除效果。
现有的釆样率偏差纠正方案至少具有如下缺陷：
在计算釆样率偏差时， 现有方案需要监测送受话端釆样时钟， 在一些情况 下需要特别的硬件设置， 比如设置高性能的 CPU, 对硬件要求较高， 另外在计 算上较为繁瑣， 占用的存储资源也较多。 而且现有方案并不直接对信号进行釆 样率偏差纠正， 而只是将釆样率偏差传递至回声消除系统由回声消除系统进行 调节处理， 这种由回声消除系统调节的方式， 增加了回声消除系统的负担， 影 响了回声消除的效果。
发明内容 本发明提供了一种送受话端釆样率偏差纠正方法和系统， 以解决现有方案 要么对硬件设置要求较高， 要么计算繁瑣、 适用范围较窄的问题以及现有方案 中不直接对信号进行釆样率偏差纠正所导致回声消除系统负担较重的问题。
为达到上述目的， 本发明实施例釆用了如下技术方案：
本发明实施例提供了一种送受话端釆样率偏差纠正方法， 所述方法包括： 根据送受话端信号计算各釆样时刻的受话端信号相对于送话端信号的传递 函数； 利用所述传递函数获取各釆样时刻送受话端的传输时延； 利用所述传输 时延和传输时延与釆样率偏差之间的线性关系， 釆用参数拟合方式得到各釆样 时刻送受话端的釆样率偏差； 根据所述釆样率偏差调整各釆样时刻送话端信号 或受话端信号的釆样率， 实现釆样率纠正， 并将纠正后釆样率相同的送话端信 号和受话端信号输入回声消除系统， 以用于回声消除系统直接利用纠正后釆样 率相同的送话端信号和受话端信号进行回声消除。 本发明实施例还提供了一种送受话端釆样率偏差纠正系统，所述系统包括： 时延估计器、 釆样率偏差估计器和釆样率调整器， 所述时延估计器的输入端接 入送话端信号和受话端信号， 所述时延估计器的输出端连接至釆样率偏差估计 器的输入端， 所述釆样率偏差估计器的输出端连接至所述釆样率调整器的输入 端， 所述釆样率调整器的输入端还接入送话端信号或者受话端信号， 所述釆样 率调整器的输出端连接至回声消除系统，
所述时延估计器， 用于根据送受话端信号计算各釆样时刻的受话端信号相 对于送话端信号的传递函数； 利用所述传递函数获取各釆样时刻送受话端的传 输时延；
所述釆样率偏差估计器， 用于利用所述传输时延和传输时延与釆样率偏差 之间的线性关系， 釆用参数拟合方式得到各釆样时刻送受话端的釆样率偏差； 所述釆样率调整器， 用于根据所述釆样率偏差调整各釆样时刻送话端信号 或受话端信号的釆样率， 实现釆样率纠正， 并将纠正后釆样率相同的送话端信 号和受话端信号输入回声消除系统， 以用于回声消除系统直接利用纠正后釆样 率相同的送话端信号和受话端信号进行回声消除。
本发明实施例的有益效果是：
本发明实施例利用传输时延与釆样率偏差之间具有线性关系的特点， 釆用 基于送受话端信号得出送受话端之间的传输时延， 并参数拟合出送受话端之间 釆样率偏差的技术手段， 能够实时得到高精度的釆样率偏差， 且无需额外的硬 件开销， 计算方法简单， 降低了系统成本。 进一步的， 由于本方案釆用了在回 声消除操作之前进行釆样率偏差纠正的技术手段， 降低了回声消除系统的负担， 提高了回声消除的质量。 附图说明 图 1为本发明实施例一提供的一种送受话端釆样率偏差纠正方法流程图； 图 2A 为本发明实施例提供的釆样率偏差恒定时传输时延和釆样时刻的关 系示意图；
图 2B为本发明实施例提供的釆样率偏差变化时传输时延和釆样时刻的关系 示意图；
图 3为本发明实施例二提供的一种釆样率偏差纠正系统的结构示意图； 图 4为本发明实施例二提供的又一种釆样率偏差纠正系统的结构示意图； 图 5为本发明实施例提供的釆样率纠正前后的回声消除效果实验结果图。 具体实施方式 为使本发明的目的、 技术方案和优点更加清楚， 下面将结合附图对本发明 实施方式作进一步地详细描述。
参见图 1 , 本发明实施例一提供的一种送受话端釆样率偏差纠正方法， 该方 法包括：
11 : 根据送受话端信号计算各釆样时刻的受话端信号相对于送话端信号的 传递函数；
12: 利用所述传递函数获取各釆样时刻送受话端的传输时延；
13 : 利用所述传输时延和传输时延与釆样率偏差之间的线性关系， 釆用参 数拟合方式得到各釆样时刻送受话端的釆样率偏差；
14: 根据所述釆样率偏差调整各釆样时刻送话端信号或受话端信号的釆样 率， 实现釆样率纠正， 并将纠正后釆样率相同的送话端信号和受话端信号输入 回声消除系统， 以用于回声消除系统直接利用纠正后釆样率相同的送话端信号 和受话端信号进行回声消除。
本发明实施例利用了传输时延与釆样率偏差之间具有线性关系的特点， 在 此通过如下分析说明利用上述特点可以纠正釆样率偏差的原理：
若送受话端存在釆样率偏差， 令 n为釆样时刻， 则受话端和送话端之间的 相对传输时延是釆样时刻的线性函数， 该线性函数可以表示如下：
^ FsS - FsR
dFs
FsR
其中， 为受话端釆样频率， 为送话端釆样频率， ^¾为釆样率偏差。 将釆样时刻 n时送受话端的传输时延表示为 ) W , 则 ) W和 n符合如下线 性关系：
D [n] = n - dFs + c
其中， C为常数， 由传输环境决定。
参见图 2A, 示出了釆样率偏差恒定时， 和 n的关系示意图， 图中的横 坐标为釆样时刻， 纵坐标为传输时延， 图中直线的斜率即为釆样率偏差， 当釆 样率偏差出现变化时， 斜率也会发生变化。 图 2B中示出了釆样率偏差变化（非 恒定） 时 )[«]和 n的关系示意图。
由上可知，如果能求出 )[«] ,则从 )[«]和 n能够估计得到 dFs ,并根据^ ¾纠 正送话端信号或受话端信号。 如果估计是实时的， 则当釆样率偏差出现变化时， 可以跟踪并适应变化， 从而能够实现在线实时地纠正送受话端的釆样率偏差。
本发明实施例利用传输时延与釆样率偏差之间具有线性关系的特点， 釆用 基于送受话端信号得出送受话端之间的传输时延， 并参数拟合出送受话端之间 釆样率偏差的技术手段， 能够实时得到高精度的釆样率偏差， 且无需额外的硬 件开销， 计算方法简单， 降低了系统成本。 进一步的， 由于本方案釆用了在回 声消除操作之前进行釆样率偏差纠正的技术手段， 降低了回声消除系统的负担，
提高回声消除的质量。 在图 1所示的实施例的基础上， 进一步的， 本实施例中上述步骤 11具体包 括： 对各釆样时刻的每个当前釆样时刻， 利用当前釆样时刻的送话端信号和当 前釆样时刻之前预定数量的送话端信号生成当前釆样时刻的送话端数据帧； 利 用当前釆样时刻的受话端信号和当前釆样时刻之前预定数量的受话端信号生成 当前釆样时刻的受话端数据帧； 利用当前釆样时刻的所述送话端数据帧和受话 端数据帧计算当前釆样时刻的受话端信号相对于送话端信号的传递函数。
进一步的， 在步骤 11中， 通过如下公式， 利用所述当前釆样时刻的送话端 数据帧和受话端数据帧计算当前釆样时刻的受话端信号相对于送话端信号的传 递函数：
= iffl {H)
其中， h为传递函数， μ]为当前釆样时刻 n的受话端数据帧 «]的频域形 式， ； τμ]为当前釆样时刻 n的送话端数据帧 的频域形式， H为传递函数 h的 频域形式， 为 μ]的共轭， E (.)表示求期望运算， t(.)表示反傅里叶变换。
进一步的， 上述步骤 12具体包括： 对各釆样时刻的每个当前釆样时刻， 选 取当前釆样时刻的传递函数的绝对值的最大值所对应的时间点， 作为当前釆样 时刻送受话端的传输时延估计值； 根据传输时延估计值得到当前釆样时刻的传 输时延， 例如， 将得到的传输时延估计值直接作为所使用的当前釆样时刻送受 话端的传输时延。
进一步的， 上述步骤 13具体包括： 对各釆样时刻的每个当前釆样时刻， 利 用当前釆样时刻送受话端的传输时延和当前釆样时刻之前预定数量的送受话端 的传输时延生成当前釆样时刻的传输时延数据帧； 根据传输时延与釆样率偏差 之间的线性关系将所述传输时延数据帧中的各元素相对于各釆样时刻进行参数 拟合， 得到当前釆样时刻送受话端的釆样率偏差。
进一步的， 上述步骤 14具体包括： 对各釆样时刻的每个当前釆样时刻， 根 据当前釆样时刻送受话端的釆样率偏差釆用内插方式对送话端信号重新进行釆 样， 得到当前釆样时刻下釆样率与受话端信号釆样率相同的送话端信号；
或者， 根据当前釆样时刻送受话端的釆样率偏差釆用内插方式对受话端信 号重新进行釆样， 得到当前釆样时刻下釆样率与送话端信号釆样率相同的受话 端信号。 上述内插方式可以为多项式内插方式或线性内插方式等。 本发明实施例二结合具体的实现器件来说明本实施例所提供的送受话端 釆样率偏差纠正方案。
参见图 3 , 示出了本实施例提供的一种釆样率偏差纠正系统， 该系统包括 时延估计器 1、 釆样率偏差估计器 2和釆样率调整器 3 , 该系统中的受话端信 号和送话端信号都是以数字方式传输的， 所以相应的， 在送话端还设置有模数 转换器， 以将釆集到的送话端的模拟信号 y ( t )转换为数字信号 y[n]。 送话端 (数字 )信号 y[n]和受话端 （数字 )信号 x[n]接入时延估计器 1的输入端， 时 延估计器 1的输出端连接至釆样率偏差估计器 2的输入端， 从时延估计器 1的 输出端输出的传输时延^ ["] 被传输至釆样率偏差估计器 2的输入端。
釆样率偏差估计器 2的输出端连接至所述釆样率调整器 3的输入端， 从釆 样率偏差估计器 2的输出端输出的釆样率偏差 被传输至釆样率调整器 3 的输入端， 釆样率调整器 3的输出端连接至回声消除系统。
应当注意到的是， 图 3所示的场景中利用得到的釆样率偏差对送话端信号 进行纠正得到釆样率相同的送受话端信号， 所以， 该场景下釆样率调整器的输 入端还接入送话端信号， 纠正后的送话端信号 和受话端信号 x[n]被传输至 回声消除系统的输入端。 图 3所示的示例中， 利用釆样率调整器对送话端信号的釆样频率进行了纠 正， 一种可选的方式中， 也可以利用釆样率调整器对受话端信号的釆样频率进 行纠正， 本实施例中主要以前者的情况为例进行说明。 对于利用得到的釆样率 偏差对受话端信号进行纠正的场景， 釆样率调整器的输入端不需接入送话端信 号， 而需要接入受话端信号， 该场景下将送话端信号 y[n]和纠正后的受话端信 号 c'[«] ]传输至回声消除系统的输入端。
上述时延估计器 1 ,用于根据送受话端信号计算各釆样时刻的受话端信号相 对于送话端信号的传递函数； 利用所述传递函数获取各釆样时刻送受话端的传 输时延；
上述釆样率偏差估计器 2,用于利用所述传输时延和传输时延与釆样率偏差 之间的线性关系， 釆用参数拟合方式得到各釆样时刻送受话端的釆样率偏差； 上述釆样率调整器 3 ,用于根据所述釆样率偏差调整各釆样时刻送话端信号 或受话端信号的釆样率， 实现釆样率偏差纠正， 并将纠正后釆样率相同的送话 端信号和受话端信号输入回声消除系统进行回声消除。 由上可见， 本实施例对送话端信号或受话端信号的釆样率纠正之后再送入 回声消除系统， 即回声消除系统的输入量为具有相同釆样频率的送受话端信号， 本实施例釆用在得到具有相同釆样频率的送受话端信号再执行回声消除的方 式， 有助于提高回声消除的效果， 而且减轻了位于数据处理后端的回声消除系 统的负担。
参见图 4, 示出了图 3中各器件的具体结构， 利用上述器件执行釆样率偏差 纠正的操作主要包括如下三部分：
一、 时延估计
本部分的操作主要由时延估计器实现， 参见图 4, 送话端数字信号 y[n]和 受话端数字信号 x[n]经过緩存器， 分别形成送话端数据帧 和受话端数据帧 χ [η] , 表示如下：
其中， n表示釆样时刻， L表示数据帧的长度， 即数据帧中元素的个数。 时延估计器 1包括受话端緩存器 11、 送话端緩存器 12、传递函数估计器 13 和时延计算器 14。
送话端緩存器 12, 用于緩存各釆样时刻的送话端信号。 受话端緩存器 11 , 用于緩存各釆样时刻的受话端信号；
上述传递函数估计器 13 , 用于对各釆样时刻的每个当前釆样时刻， 利用送 话端緩存器中当前釆样时刻的送话端信号和当前釆样时刻之前预定数量的送话 端信号生成当前釆样时刻的送话端数据帧； 利用受话端緩存器中当前釆样时刻
的受话端信号和当前釆样时刻之前预定数量的受话端信号生成当前釆样时刻的 受话端数据帧； 以及利用当前釆样时刻的所述送话端数据帧和受话端数据帧计 算当前釆样时刻的受话端信号相对于送话端信号的传递函数；
上述的预定数量为 L-1 , 即时延估计器 1的传递函数估计器 13利用当前釆 样时刻的送话端信号 y[n]和当前釆样时刻之前 L-1 个送话端信号生成当前釆样 时刻的送话端数据帧 以及， 利用当前釆样时刻的受话端信号 x[n]和当前釆 样时刻之前 L-1个受话端信号生成当前釆样时刻的受话端数据帧 , L的具体 数值与系统的输出时延限制有关， 如 L可以取 256或 512等。
将数据帧 和 送至传递函数估计器 13 的输入端， 计算 到 «]的 传递函数 h。 传递函数估计器 13 , 通过如下计算方式， 利用当前釆样时刻的所 述送话端数据帧和受话端数据帧计算当前釆样时刻的受话端信号相对于送话端 信号的传递函数， 传递函数计算方式可以釆用互功率谱和自功率语相除的方式， 具体公式如下：
其中， h为传递函数， μ]为当前釆样时刻 n的受话端数据帧 j [«]的频域形 式， ； τμ]为当前釆样时刻 n的送话端数据帧 的频域形式， H为传递函数 h的 频域形式， 为 μ]的共轭， E (.)表示求期望运算， t(.)表示反傅里叶变换。
将传递函数估计器 13 计算出的传递函数输入时延计算器 14, 时延计算器 14对各釆样时刻的每个当前釆样时刻， 选取当前釆样时刻的传递函数的绝对值 的最大值所对应的时间点， 作为当前釆样时刻送受话端的传输时延估计值； 以 及， 根据所述传输时延估计值得到所使用的当前釆样时刻送受话端的传输时延。 本实施例中选取 h绝对值的最大值位置所对应的 h参数 (时间参数） 的数 值作为送受话端信号时延估计值 。
由上， 通过时延估计器 1的计算， 对于每个釆样时刻 n都能计算得到相应 的送受话端传输时延 D£ W。
注： 由于本实施例釆用了利用当前釆样时刻及当前釆样时刻之间的信号来 估计传输时延的方式， 对于初始釆样时刻， 可以将该初始釆样时刻之前的信号 的数值釆用默认数值（如 0 )。
二、 采样率偏差估计
本部分的操作主要由釆样率偏差估计器实现， 参见图 4, 釆样率偏差估计器 包括时延緩存器 21和时延釆样率偏差拟合器 22。 将计算出的传输时延 [«]送 入釆样率偏差估计器的时延緩存器 21 , 形成传输时延数据帧 相对应的釆 样时刻所形成的数据帧记录为 ， 则有：
DE [n] = (DE [n - M + \] · · · DE [n - \] DE [n])
n = {n - M + \ ·■■ n - \ n)
其中， M是数据帧的长度。 M体现了观测时间的长短， 本实施例中， 观测 时间的增长能够提高拟合的精度。
由上， 釆样率偏差估计器 2中的时延緩存器 21 , 用于緩存各釆样时刻送受 话端的传输时延。 时延釆样率偏差拟合器 22, 用于对各釆样时刻的每个当前釆 样时刻， 利用时延緩存器中当前釆样时刻送受话端的传输时延和当前釆样时刻 之前预定数量的送受话端的传输时延生成当前釆样时刻的传输时延数据帧； 以 及， 根据传输时延与釆样率偏差之间的线性关系将所述传输时延数据帧中的各 元素相对于各釆样时刻进行参数拟合， 得到当前釆样时刻送受话端的釆样率偏 差。
即时延釆样率偏差拟合器 22, 利用当前釆样时刻 n送受话端的传输时延 DE [n]和当前釆样时刻之前 M-1个送受话端的传输时延生成当前釆样时刻的传输 时延数据帧 然后， 釆用时延釆样率偏差拟合器 22根据传输时延与釆样 率偏差之间的线性关系将所述传输时延数据帧中的各元素相对于各釆样时刻进 行参数拟合， 得到当前釆样时刻送受话端的釆样率偏差。 将 和 送入时延 釆样率偏差拟合器， 估计 对 的斜率， 此斜率就是估计出的釆样率偏差 dFsE。
拟合方式可以釆用最小二乘拟合、 最大似然拟合或者其他参数拟合方式。 当釆用最小二乘拟合时， 具体的计算公式可以表示如下：
∑ {k - ^ ^) (DE [k] - E (DE [n]))
dFsE = M+1、 —— J—
Var[n) 其中， 表示釆样率偏差， 表示求期望运算， ^(.)为求方差运算。 最小二乘拟合方式以及其他参数拟合方式的估计精度都会随 M的增加而提 高， 也即是说： 当观测时间 M延长时， 的估计精度会提高。
通过上述操作， 对于每个釆样时刻 n都可以实时计算得到釆样率偏差。 三、 采样率调整
本部分的操作主要由釆样率调整器 3实现， 釆样率调整器 3根据所述釆样 率偏差纠正各釆样时刻送话端信号或受话端信号的釆样率， 以使回声消除系统 利用纠正后釆样率相同的送话端信号和受话端信号进行回声消除。
具体的， 在每个釆样时刻， 釆样率调整器 3根据当前釆样时刻的釆样率偏 差对当前釆样时刻的送话端信号或受话端信号重新进行釆样， 得到当前釆样时 刻下釆样率相同的送话端信号和受话端信号。
进一步的， 上述釆样率调整器 3包括重釆样緩存器 31 , 当所述釆样率调整器 3的输入端接入送话端信号时， 如图 4中所示的场景， 该场景下所述釆样率调整 器 3包括緩存各釆样时刻的送话端信号的重釆样緩存器 31 , 这时， 所述釆样率 调整器 3 , 具体用于对各釆样时刻的每个当前釆样时刻， 根据当前釆样时刻送受 话端的釆样率偏差， 釆用内插方式对重釆样緩存器 31中的送话端信号重新进行 釆样， 得到当前釆样时刻下釆样率与受话端信号釆样率相同的送话端信号； 当釆样率调整器 3的输入端接入受话端信号时， 釆样率调整器 3 包括緩存 各釆样时刻的受话端信号的重釆样緩存器 31 , 这时， 釆样率调整器 3 , 具体用 于对各釆样时刻的每个当前釆样时刻， 根据当前釆样时刻送受话端的釆样率偏 差， 釆用内插方式对重釆样緩存器 31中的受话端信号重新进行釆样， 得到当前 釆样时刻下釆样率与送话端信号釆样率相同的受话端信号。
由上可见， 虽然时延估计器 1需要同时利用送受话端信号， 釆样率调整器 3 需要利用送话端信号或者受话端信号， 然而， 两者利用的送话端信号可能并不 相同， 两者利用的受话端信号可能也不相同， 为避免时延估计器 1 和釆样率调 整器 3相互影响， 提高数据处理速度， 本实施例釆用分别设置针对时延估计器 1 的受话端緩存器和送话端緩存器， 以及针对釆样率调整器 3 的送话端重釆样緩 存器或受话端重釆样緩存器的方式。
重釆样可以釆用多项式内插、 线性内插或其它通用的重釆样方式。 以线性 内插方式为例， 如果对送话端信号进行重釆样， 如图 4 中所示的场景， 则重釆 样后的送话端信号可以表示如下：
_ '["] = · (1 + dFsE ) + 1 -[« · (ΐ + dFsE )^ y -(1 + dFsE )J
+ · (1 + dFsE
其中， L.」表示向下取整运算， 表示重釆样后的釆样时刻 n的送话端信 号。
如果对受话端信号进行重釆样， 则重釆样后的受话端信号可以表示如下： l + dFsE ) + l - [
其中， L.」表示向下取整运算， x'W表示重釆样后的釆样时刻 n的受话端信 号。
至此就实现了在线釆样率偏差纠正， 将纠正后的釆样率相同的送受话端信 号送入回声消除系统进行回声消除。
本实施例中的器件不需要额外的特殊设置， 例如， 上述时延估计器 1 可以 由緩存器、 传递函数估计器 13和时延计算器 14实现， 而传递函数估计器 13 可以由乘法器、 积分器、 除法器和反傅里叶变换器实现。 时延计算器 14 可以 由乘法器、 除法器和比较器实现等等。
由上所述， 本实施例基于送受端信号进行釆样率偏差的纠正， 无需监测送 受话端釆样时钟， 从而无需额外的硬件设置， 降低了对硬件性能的要求， 节省 了系统成本。
并且， 本发明实施例利用传输时延与釆样率偏差之间具有线性关系的特点， 釆用基于送受话端信号得出送受话端之间的传输时延， 并参数拟合出送受话端 之间釆样率偏差的技术手段， 能够实时得到高精度的釆样率偏差， 且无需额外 的硬件开销， 计算方法简单， 降低了系统成本。 进一步的， 由于本方案釆用了 在回声消除操作之前进行釆样率偏差纠正的技术手段， 降低了回声消除系统的 负担， 提高回声消除的质量。 下面结合实验结果图说明本发明实施例的有益效果。 图 5是釆样率纠正前 后的回声消除效果实验结果图。送受话端的期望釆样率为 16000Hz,存在的釆样 率偏差为 0.9Hz, 从 29秒开始做釆样率偏差纠正。 在 29秒之前将具有釆样率偏 差的送受话端信号输入至回声消除系统， 在 29秒之后将釆用本方案得到的具有 相同釆样率的送受话端信号输入至回声消除系统。
图 5中的纵坐标为 RMS (均方才艮误差） Power (能量）， 横坐标为时间， 图 中短划线为送话端 Mic (麦克）信号的能量曲线， 实线为回声消除系统输出的残 留回声能量曲线， 两条曲线的差值即为回声抑制量。 可以看出， 在未经釆样率 偏差纠正时（29秒之前）， 回声抑制量只能达到 21dB, 经过纠正后（29秒之后） 回声抑制量可以超过 45dB, 从而验证了本方案能够显著提高回声消除的效果。 以上所述仅为本发明的较佳实施例而已， 并非用于限定本发明的保护范围。 凡在本发明的精神和原则之内所作的任何修改、 等同替换、 改进等， 均包含在 本发明的保护范围内。
Claims
1、 一种送受话端釆样率偏差纠正方法， 其中， 所述方法包括：
根据送受话端信号计算各釆样时刻的受话端信号相对于送话端信号的传递 函数；
利用所述传递函数获取各釆样时刻送受话端的传输时延；
利用所述传输时延和传输时延与釆样率偏差之间的线性关系， 釆用参数拟 合方式得到各釆样时刻送受话端的釆样率偏差；
根据所述釆样率偏差调整各釆样时刻送话端信号或受话端信号的釆样率， 实现釆样率纠正， 并将纠正后釆样率相同的送话端信号和受话端信号输入回声 消除系统进行回声消除。
2、 根据权利要求 1所述的方法， 其中， 所述根据送受话端信号计算各釆样 时刻的受话端信号相对于送话端信号的传递函数包括：
对各釆样时刻的每个当前釆样时刻，
利用当前釆样时刻的送话端信号和当前釆样时刻之前预定数量的送话端信 号生成当前釆样时刻的送话端数据帧； 同时利用当前釆样时刻的受话端信号和 当前釆样时刻之前预定数量的受话端信号生成当前釆样时刻的受话端数据帧； 通过如下公式， 利用当前釆样时刻的所述送话端数据帧和受话端数据帧计 算当前釆样时刻的受话端信号相对于送话端信号的传递函数：
= iffl {H)
其中， h为传递函数， μ]为当前釆样时刻 n的受话端数据帧 j [«]的频域形 式， ； τμ]为当前釆样时刻 n的送话端数据帧 的频域形式， H为传递函数 h的 频域形式， 为 μ]的共轭， E (.)表示求期望运算， t(.)表示反傅里叶变换。
3、 根据权利要求 1所述的方法， 其中， 所述利用所述传递函数获取各釆样 时刻送受话端的传输时延包括：
对各釆样时刻的每个当前釆样时刻，
选取当前釆样时刻的传递函数的绝对值的最大值所对应的时间点， 作为当
前釆样时刻送受话端的传输时延估计值；
根据所述传输时延估计值得到当前釆样时刻送受话端的传输时延。
4、 根据权利要求 1所述的方法， 其中， 所述利用所述传输时延和传输时延 与釆样率偏差之间的线性关系， 釆用参数拟合方式得到各釆样时刻送受话端的 釆样率偏差包括：
对各釆样时刻的每个当前釆样时刻，
利用当前釆样时刻送受话端的传输时延和当前釆样时刻之前预定数量的送 受话端的传输时延生成当前釆样时刻的传输时延数据帧；
根据传输时延与釆样率偏差之间的线性关系将所述传输时延数据帧中的各 元素相对于各釆样时刻进行参数拟合， 得到当前釆样时刻送受话端的釆样率偏 差。
5、 根据权利要求 1所述的方法， 其中， 所述根据所述釆样率偏差纠正各釆 样时刻送话端信号或受话端信号的釆样率包括：
对各釆样时刻的每个当前釆样时刻，
根据当前釆样时刻送受话端的釆样率偏差釆用内插方式对送话端信号重新 进行釆样， 得到当前釆样时刻下釆样率与受话端信号釆样率相同的送话端信号； 或者， 根据当前釆样时刻送受话端的釆样率偏差釆用内插方式对受话端信 号重新进行釆样， 得到当前釆样时刻下釆样率与送话端信号釆样率相同的受话 端信号。
6、 一种送受话端釆样率偏差纠正系统， 其中， 所述系统包括时延估计器、 釆样率偏差估计器和釆样率调整器，
所述时延估计器的输入端接入送话端信号和受话端信号， 所述时延估计器 的输出端连接至釆样率偏差估计器的输入端， 所述釆样率偏差估计器的输出端 连接至所述釆样率调整器的输入端， 所述釆样率调整器的输入端还接入送话端 信号或者受话端信号， 所述釆样率调整器的输出端连接至回声消除系统，
所述时延估计器， 用于根据送受话端信号计算各釆样时刻的受话端信号相 对于送话端信号的传递函数， 并利用所述传递函数获取各釆样时刻送受话端的
传输时延；
所述釆样率偏差估计器， 用于利用所述传输时延和传输时延与釆样率偏差 之间的线性关系， 釆用参数拟合方式得到各釆样时刻送受话端的釆样率偏差； 所述釆样率调整器， 用于根据所述釆样率偏差调整各釆样时刻送话端信号 或受话端信号的釆样率， 实现釆样率纠正， 并将纠正后釆样率相同的送话端信 号和受话端信号输入回声消除系统进行回声消除。
7、根据权利要求 6所述的系统， 其中， 所述时延估计器包括送话端緩存器、 受话端緩存器和传递函数估计器，
所述送话端緩存器， 用于緩存各釆样时刻的送话端信号；
所述受话端緩存器， 用于緩存各釆样时刻的受话端信号；
所述传递函数估计器， 用于对各釆样时刻的每个当前釆样时刻， 利用所述 送话端緩存器中当前釆样时刻的送话端信号和当前釆样时刻之前预定数量的送 话端信号生成当前釆样时刻的送话端数据帧； 同时利用所述受话端緩存器中当 前釆样时刻的受话端信号和当前釆样时刻之前预定数量的受话端信号生成当前 釆样时刻的受话端数据帧； 以及， 通过如下公式， 利用当前釆样时刻的所述送 话端数据帧和受话端数据帧计算当前釆样时刻的受话端信号相对于送话端信号 的传递函数：
= iffl {H)
其中， h为传递函数， μ]为当前釆样时刻 n的受话端数据帧 的频域形 式， ； τμ]为当前釆样时刻 n的送话端数据帧 的频域形式， H为传递函数 h的 频域形式， 为 μ]的共轭， E (.)表示求期望运算， t(.)表示反傅里叶变换。
8、根据权利要求 7所述的系统， 其中， 所述时延估计器还包括时延计算器， 所述时延计算器， 用于对各釆样时刻的每个当前釆样时刻， 选取当前釆样 时刻的传递函数的绝对值的最大值所对应的时间点， 作为当前釆样时刻送受话 端的传输时延估计值； 根据所述传输时延估计值得到当前釆样时刻送受话端的 传输时延。
9、 根据权利要求 6所述的系统， 其中， 所述釆样率偏差估计器包括时延 緩存器和时延釆样率偏差拟合器，
所述时延緩存器， 用于緩存各釆样时刻送受话端的传输时延；
所述时延釆样率偏差拟合器， 用于对各釆样时刻的每个当前釆样时刻， 利 用所述时延緩存器中当前釆样时刻送受话端的传输时延和当前釆样时刻之前预 定数量的送受话端的传输时延生成当前釆样时刻的传输时延数据帧； 以及， 根 据传输时延与釆样率偏差之间的线性关系将所述传输时延数据帧中的各元素相 对于各釆样时刻进行参数拟合， 得到当前釆样时刻送受话端的釆样率偏差。
10、 根据权利要求 6所述的系统， 其中， 所述釆样率调整器包括重釆样緩 存器；
当所述釆样率调整器的输入端接入送话端信号时， 所述重釆样緩存器用于 緩存各釆样时刻的送话端信号， 所述釆样率调整器， 具体用于对各釆样时刻的 每个当前釆样时刻， 根据当前釆样时刻送受话端的釆样率偏差， 釆用内插方式 对所述重釆样緩存器中的送话端信号重新进行釆样， 得到当前釆样时刻下釆样 率与受话端信号釆样率相同的送话端信号；
当所述釆样率调整器的输入端接入受话端信号时， 所述重釆样緩存器用于 緩存各釆样时刻的受话端信号， 所述釆样率调整器， 具体用于对各釆样时刻的 每个当前釆样时刻， 根据当前釆样时刻送受话端的釆样率偏差， 釆用内插方式 对所述重釆样緩存器中的受话端信号重新进行釆样， 得到当前釆样时刻下釆样 率与送话端信号釆样率相同的受话端信号。
Priority Applications (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2014537488A JP5629408B1 (ja) | 2012-07-06 | 2013-05-30 | 送受話端サンプリングレート偏差の補正方法及びシステム |
| US14/351,225 US9065896B2 (en) | 2012-07-06 | 2013-05-30 | Method and system for sampling rate mismatch correction of transmitting and receiving terminals |
| KR1020147010046A KR101466543B1 (ko) | 2012-07-06 | 2013-05-30 | 송신기단과 수신기단의 샘플링 레이트 편차를 정정하기 위한 방법 및 시스템 |
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210233695.9 | 2012-07-06 | ||
| CN201210233695.9A CN102780821B (zh) | 2012-07-06 | 2012-07-06 | 一种送受话端采样率偏差纠正方法和系统 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2014005473A1 true WO2014005473A1 (zh) | 2014-01-09 |
Family
ID=47125573
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2013/076487 WO2014005473A1 (zh) | 2012-07-06 | 2013-05-30 | 一种送受话端采样率偏差纠正方法和系统 |
Country Status (5)
| Country | Link |
|---|---|
| US (1) | US9065896B2 (zh) |
| JP (1) | JP5629408B1 (zh) |
| KR (1) | KR101466543B1 (zh) |
| CN (1) | CN102780821B (zh) |
| WO (1) | WO2014005473A1 (zh) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8750494B2 (en) * | 2011-08-17 | 2014-06-10 | Alcatel Lucent | Clock skew compensation for acoustic echo cancellers using inaudible tones |
| CN102780821B (zh) | 2012-07-06 | 2014-08-13 | 歌尔声学股份有限公司 | 一种送受话端采样率偏差纠正方法和系统 |
| CN108605304B (zh) * | 2016-02-15 | 2021-05-14 | 瑞典爱立信有限公司 | 采用降低的采样率的NB-IoT装置中的下行链路时间跟踪 |
| CN108605031A (zh) | 2016-02-15 | 2018-09-28 | 瑞典爱立信有限公司 | 以最小采样率操作的nb-iot接收器 |
| CN105872156B (zh) * | 2016-05-25 | 2019-02-12 | 腾讯科技（深圳）有限公司 | 一种回声时延跟踪方法及装置 |
| CN106210371B (zh) * | 2016-08-31 | 2018-09-18 | 广州视源电子科技股份有限公司 | 一种回声时延的确定方法、装置及智能会议设备 |
| CN110992973A (zh) * | 2019-11-29 | 2020-04-10 | 维沃移动通信有限公司 | 一种信号时延的确定方法和电子设备 |
| CN114141224B (zh) * | 2021-11-30 | 2023-06-09 | 北京百度网讯科技有限公司 | 信号处理方法和装置、电子设备、计算机可读介质 |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100086122A1 (en) * | 2008-10-02 | 2010-04-08 | Oki Electric Industry Co., Ltd. | Echo canceller and echo cancelling method and program |
| CN101819768A (zh) * | 2009-02-25 | 2010-09-01 | 富士通株式会社 | 噪声抑制装置和噪声抑制方法 |
| CN102387272A (zh) * | 2011-09-09 | 2012-03-21 | 南京大学 | 一种回声抵消系统中残留回声的抑制方法 |
| CN102780821A (zh) * | 2012-07-06 | 2012-11-14 | 歌尔声学股份有限公司 | 一种送受话端采样率偏差纠正方法和系统 |
Family Cites Families (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6167133A (en) * | 1997-04-02 | 2000-12-26 | At&T Corporation | Echo detection, tracking, cancellation and noise fill in real time in a communication system |
| US6563863B1 (en) * | 1998-06-22 | 2003-05-13 | Ati International Srl | Computer modem |
| JP3673727B2 (ja) | 2000-11-22 | 2005-07-20 | 日本電信電話株式会社 | 反響消去方法、その装置、そのプログラム及びその記録媒体 |
| JP4492953B2 (ja) * | 2004-12-21 | 2010-06-30 | ルネサスエレクトロニクス株式会社 | キャンセラ装置及びデータ伝送システム |
| NO327377B1 (no) * | 2007-12-18 | 2009-06-22 | Tandberg Telecom As | Fremgangsmate og system for klokkedriftskompensering |
| JP4834046B2 (ja) * | 2008-08-27 | 2011-12-07 | 日本電信電話株式会社 | エコー消去装置、エコー消去方法、エコー消去プログラム、記録媒体 |
| JP2010245629A (ja) * | 2009-04-01 | 2010-10-28 | Mitsubishi Electric Corp | 適応等化器および適応等化方法 |
| CN101645273B (zh) * | 2009-07-10 | 2012-02-01 | 中国科学院声学研究所 | 一种采样率差异估计与校正系统及其处理方法 |
| JP5421877B2 (ja) * | 2010-08-25 | 2014-02-19 | 日本電信電話株式会社 | エコー消去方法、エコー消去装置及びエコー消去プログラム |
| CN102457632B (zh) * | 2011-12-29 | 2014-07-30 | 歌尔声学股份有限公司 | 一种多受话端回声消除方法 |
| CN202949477U (zh) * | 2012-07-06 | 2013-05-22 | 歌尔声学股份有限公司 | 一种送受话端采样率偏差纠正系统 |
| JPWO2014115290A1 (ja) * | 2013-01-25 | 2017-01-26 | 株式会社日立製作所 | 信号処理装置・音響処理システム |
-
2012
- 2012-07-06 CN CN201210233695.9A patent/CN102780821B/zh active Active
-
2013
- 2013-05-30 US US14/351,225 patent/US9065896B2/en active Active
- 2013-05-30 KR KR1020147010046A patent/KR101466543B1/ko active IP Right Grant
- 2013-05-30 WO PCT/CN2013/076487 patent/WO2014005473A1/zh active Application Filing
- 2013-05-30 JP JP2014537488A patent/JP5629408B1/ja active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100086122A1 (en) * | 2008-10-02 | 2010-04-08 | Oki Electric Industry Co., Ltd. | Echo canceller and echo cancelling method and program |
| CN101819768A (zh) * | 2009-02-25 | 2010-09-01 | 富士通株式会社 | 噪声抑制装置和噪声抑制方法 |
| CN102387272A (zh) * | 2011-09-09 | 2012-03-21 | 南京大学 | 一种回声抵消系统中残留回声的抑制方法 |
| CN102780821A (zh) * | 2012-07-06 | 2012-11-14 | 歌尔声学股份有限公司 | 一种送受话端采样率偏差纠正方法和系统 |
Also Published As
| Publication number | Publication date |
|---|---|
| JP5629408B1 (ja) | 2014-11-19 |
| KR20140064963A (ko) | 2014-05-28 |
| CN102780821A (zh) | 2012-11-14 |
| JP2014533014A (ja) | 2014-12-08 |
| US9065896B2 (en) | 2015-06-23 |
| CN102780821B (zh) | 2014-08-13 |
| US20140233723A1 (en) | 2014-08-21 |
| KR101466543B1 (ko) | 2014-11-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2014005473A1 (zh) | 一种送受话端采样率偏差纠正方法和系统 | |
| CN110197669B (zh) | 一种语音信号处理方法及装置 | |
| JP2002529753A (ja) | 改善された信号定位装置 | |
| JP5257366B2 (ja) | 雑音抑圧装置、雑音抑圧制御装置、雑音抑圧方法及び雑音抑圧プログラム | |
| WO2011006369A1 (zh) | 一种改进的离散余弦变换域音频丢帧补偿器和补偿方法 | |
| JP6160403B2 (ja) | エコー抑圧装置及びエコー抑圧プログラム | |
| JP6023884B2 (ja) | 音声パケット損失を補償する方法及び装置 | |
| CN109727607B (zh) | 时延估计方法、装置及电子设备 | |
| JP2011512698A (ja) | クロックドリフト補償の方法とシステム | |
| JP6295722B2 (ja) | エコー抑圧装置、プログラム及び方法 | |
| US10341766B1 (en) | Microphone apparatus and headset | |
| WO2012092900A2 (zh) | 一种信道频偏估计方法、装置及系统 | |
| WO2014089914A1 (zh) | 一种基于双麦克的语音混响消减方法和装置 | |
| TW201021564A (en) | Single-carrier/multi-carrier community receiver | |
| WO2007068166A1 (fr) | Dispositif et procede d'elimination d'echo d'electricite | |
| US9894445B2 (en) | Method for operating a hearing device and hearing device | |
| AU2018331317B2 (en) | Selecting channel adjustment method for inter-frame temporal shift variations | |
| WO2012158168A1 (en) | Clock drift compensation method and apparatus | |
| US20140128004A1 (en) | Converting samples of a signal at a sample rate into samples of another signal at another sample rate | |
| CN202949477U (zh) | 一种送受话端采样率偏差纠正系统 | |
| CN106328160A (zh) | 一种基于双麦克的降噪方法 | |
| EP2768247B1 (en) | Method for operating a hearing device and hearing device | |
| EP2768246B1 (en) | Method for operating a hearing device and hearing device | |
| KR101583715B1 (ko) | 셋톱 박스 영상 및 음성 통화 서비스에서 적응적 필터의 반향 경로 예측을 위한 지연 시간 자동 측정 방법 및 장치 | |
| Hanshi et al. | Efficient acoustic echo cancellation joint with noise reduction framework |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 13812720 Country of ref document: EP Kind code of ref document: A1 |
|
| ENP | Entry into the national phase |
Ref document number: 2014537488 Country of ref document: JP Kind code of ref document: A |
|
| WWE | Wipo information: entry into national phase |
Ref document number: 14351225 Country of ref document: US |
|
| ENP | Entry into the national phase |
Ref document number: 20147010046 Country of ref document: KR Kind code of ref document: A |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 13812720 Country of ref document: EP Kind code of ref document: A1 |