WO2013166761A1 - 回声消除方法及装置 - Google Patents

回声消除方法及装置 Download PDF

Info

Publication number
WO2013166761A1
WO2013166761A1 PCT/CN2012/077022 CN2012077022W WO2013166761A1 WO 2013166761 A1 WO2013166761 A1 WO 2013166761A1 CN 2012077022 W CN2012077022 W CN 2012077022W WO 2013166761 A1 WO2013166761 A1 WO 2013166761A1
Authority
WO
WIPO (PCT)
Prior art keywords
signal
echo
short
far
term energy
Prior art date
Application number
PCT/CN2012/077022
Other languages
English (en)
French (fr)
Inventor
李兴波
徐发国
刘婷
马清
李晓亮
Original Assignee
中兴通讯股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 中兴通讯股份有限公司 filed Critical 中兴通讯股份有限公司
Priority to EP12876306.7A priority Critical patent/EP2849351B1/en
Priority to US14/400,289 priority patent/US9443528B2/en
Publication of WO2013166761A1 publication Critical patent/WO2013166761A1/zh

Links

Classifications

    • 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
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10K—SOUND-PRODUCING DEVICES; METHODS OR DEVICES FOR PROTECTING AGAINST, OR FOR DAMPING, NOISE OR OTHER ACOUSTIC WAVES IN GENERAL; ACOUSTICS NOT OTHERWISE PROVIDED FOR
    • G10K11/00—Methods or devices for transmitting, conducting or directing sound in general; Methods or devices for protecting against, or for damping, noise or other acoustic waves in general
    • G10K11/16—Methods or devices for protecting against, or for damping, noise or other acoustic waves in general
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B3/00—Line transmission systems
    • H04B3/02—Details
    • H04B3/20—Reducing echo effects or singing; Opening or closing transmitting path; Conditioning for transmission in one direction or the other
    • H04B3/23—Reducing echo effects or singing; Opening or closing transmitting path; Conditioning for transmission in one direction or the other using a replica of transmitted signal in the time domain, e.g. echo cancellers
    • 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

Definitions

  • BACKGROUND OF THE INVENTION 1 Field of the Invention The present invention relates to the field of communications, and in particular to an echo cancellation method and apparatus.
  • BACKGROUND OF THE INVENTION Audio quality has always been a major problem plaguing conference calls.
  • the conference terminals generally consist of two channels.
  • the downlink receiving channel decodes, reconstructs, and amplifies the voice signal transmitted by the peer end to the speaker for playback.
  • the uplink transmission channel is near.
  • the voice signal of the end user is collected through the microphone, processed by the voice, encoded, and sent to the opposite end of the conference.
  • the speaker and microphone may work at the same time, as shown in Figure 1.
  • the far-end voice signal played by the speaker may be collected by the near-end microphone and transmitted to the far-end caller through the uplink transmission channel, because the uplink transmission channel will be far away.
  • the far-end talker may hear his own speech, that is, the echo, and the larger the delay, the more obvious the echo. Therefore, echo has the greatest impact on voice calls such as VOIP.
  • echo cancellation technologies such as patent number
  • Near-end echo cancellation may be used at the unit so that the far-end participant does not hear his own voice, however, in some cases the remote unit may not have an available acoustic echo canceller to eliminate the far-end echo, In this case, the near-end participant will hear his own voice returning to the near end due to the acoustic coupling between the far-end speaker and the microphone. Therefore, near-end echo cancellation may benefit the far-end participants, but is not useful for preventing near-end participants from hearing near-end audio that is echoed back from the far end.
  • Patent Application No. CN201010287069.9 proposes a method for suppressing near-end interference at the near end to the far-end backhaul.
  • This patent application discloses the detection and suppression of backhaul audio at the near end, i.e. audio from the near end, acoustically coupled at the far end and backhauled to the near end unit is detected and suppressed at the near end of the conference. Determining the separation frequency for the near-end audio transmitted from the near-end unit and the far-end audio received at the near-end unit a first energy output and a second energy output of the band, the near-end unit compares the first energy output and the second energy output of each frequency band with each other within a certain time delay range, and detects the received on the basis of the comparison result The backend of the transmitted near-end audio in the far-end audio.
  • the near-end unit suppresses any detected back-transmitted near-end audio by muting or attenuating the far-end audio output at its speaker. That is, its core idea is: Through the double-ended speech detector unit, if the voice of the far-end participant is not detected at the near end, the near-end speaker can be muted or reduced in volume to suppress the returned near-end audio.
  • This method can effectively suppress the back-end audio of the backhaul, but the method can not adaptively suppress the back-end audio, and can not control the volume of the near-end speaker in real time; in addition, the suppression backhaul
  • the method of near-end audio is to silence or reduce the volume of the near-end speaker, which may result in discontinuity of the far-end audio, thereby reducing the voice quality of the conference and affecting the user experience. Therefore, in the above related art, when the back-end audio is suppressed, there is a discontinuity of the far-end audio due to the inability to adaptively mute or reduce the volume of the near-end speaker, thereby reducing the voice quality of the conference. A problem that affects the user experience.
  • an echo cancellation method comprising: estimating an echo path characteristic parameter of an echo signal; using a source signal of the echo signal as a reference signal, generating an echo estimation signal according to an echo path characteristic parameter; The echo estimation signal is subtracted from the speech signal.
  • the voice signal to be processed includes: a near-end voice signal and/or a voice signal transmitted from a far end.
  • the echo cancellation method further includes: acquiring a short-term energy estimate of the near-end speech signal and a short-term energy estimation of the far-end transmitted speech signal within a preset duration a value; determining a call state mode according to a short-term energy estimate of the near-end voice signal and a short-term energy estimate of the voice signal transmitted by the far-end, wherein the call state mode includes a near-end call and a far-end call.
  • determining the call state mode according to the short-term energy estimate of the near-end speech signal and the short-term energy estimate of the far-end transmitted voice signal comprises: the short-term energy estimate of the near-end speech signal is greater than the pre- In a case where a predetermined multiple of the short-term energy estimation value of the voice signal transmitted from the far end is set in the duration, the call state mode is determined to be a near-end call, and the to-be-processed voice signal is a voice signal transmitted from the far end; otherwise, the call is determined.
  • the state mode is a far-end call, and the pending voice signal is a near-end voice signal.
  • estimating an echo path characteristic parameter of the echo signal comprises: acquiring a delay of the echo signal in the to-be-processed speech signal; acquiring a power error between the previous moment and the current moment of the to-be-processed speech signal; according to the delay and the power error Estimate the echo path characteristic parameters of the echo signal.
  • the echo cancellation method further includes: determining that the power error of the to-be-processed speech signal between the previous time and the current time is greater than a preset threshold.
  • the echo path feature parameters are updated, wherein the echo path feature parameters include: an uplink echo path feature parameter and/or a downlink echo path feature parameter.
  • an echo canceling apparatus comprising: an estimating module configured to estimate an echo path characteristic parameter of an echo signal; a generating module configured to use a source signal of the echo signal as a reference signal, according to the echo The path characteristic parameter generates an echo estimation signal; and the processing module is configured to subtract the echo estimation signal from the to-be-processed speech signal.
  • the echo cancellation device further includes: an acquisition module, configured to acquire a short-term energy estimation value of the near-end speech signal and a short-term energy estimation value of the far-end transmitted speech signal within a preset duration; determining module, setting The call state mode is determined according to a short-term energy estimate of the near-end voice signal and a short-term energy estimate of the voice signal transmitted from the far-end, wherein the call state mode includes a near-end call and a far-end call.
  • an acquisition module configured to acquire a short-term energy estimation value of the near-end speech signal and a short-term energy estimation value of the far-end transmitted speech signal within a preset duration
  • determining module setting The call state mode is determined according to a short-term energy estimate of the near-end voice signal and a short-term energy estimate of the voice signal transmitted from the far-end, wherein the call state mode includes a near-end call and a far-end call.
  • the determining module comprises: a first determining unit, configured to: if the short-term energy estimation value of the near-end speech signal is greater than a preset multiple of the short-term energy estimation value of the speech signal transmitted by the far-end within the preset duration Determining that the call state mode is a near-end call, and the pending voice signal is a voice signal transmitted by the far-end; the second determining unit is configured to set the short-term energy estimate of the near-end voice signal to be less than a preset time length for the remote transmission.
  • the call state mode is determined to be a far-end call, and the pending voice signal is a near-end voice signal.
  • the estimating module includes: a first acquiring unit configured to acquire a delay of the echo signal in the to-be-processed voice signal; and a second acquiring unit configured to obtain a power error between the previous time and the current time of the to-be-processed voice signal
  • the estimation unit is configured to estimate the echo path characteristic parameters of the echo signal according to the delay and the power error.
  • the echo cancellation device further includes: an update module, configured to: when determining that the power error of the to-be-processed speech signal between the previous time and the current time is greater than a preset threshold, updating the echo path characteristic parameter, wherein, the echo
  • the path characteristic parameters include: an uplink echo path characteristic parameter and/or a downlink echo path characteristic parameter.
  • an echo estimation signal is generated according to the echo path characteristic parameter, and finally, the echo estimation signal is subtracted from the to-be-processed speech signal.
  • the echo estimation signal can be subtracted from the to-be-processed speech signal in real time and adaptively, and the echo can be eliminated.
  • the echo can be either a downlink echo or an uplink echo; and, in the present invention, The purpose of eliminating the echo is achieved by subtracting the echo estimation signal from the speech signal to be processed, thereby avoiding the discontinuity of the far-end audio by eliminating the echo of the volume of the near-end speaker or reducing the control to eliminate the echo. Problems that help improve the voice quality of the conference and improve the user experience.
  • FIG. 1 is a flow chart of an echo canceling method according to an embodiment of the present invention
  • FIG. 2 is a block diagram showing the structure of an echo canceling apparatus according to an embodiment of the present invention
  • FIG. 3 is another echo according to an embodiment of the present invention.
  • FIG. 4 is a structural block diagram of a determining module according to an embodiment of the present invention.
  • FIG. 5 is a structural block diagram of an estimating module according to an embodiment of the present invention
  • FIG. 6 is another echo canceling according to an embodiment of the present invention.
  • FIG. 7 is a functional block diagram of an echo canceling system according to an embodiment of the present invention;
  • FIG. 8 is a functional block diagram of an echo canceling apparatus according to an embodiment of the present invention;
  • FIG. 9 is an uplink echo canceling according to an embodiment of the present invention.
  • the echo cancellation method includes steps S102 to S106.
  • Step S102 Estimating an echo path characteristic parameter of the echo signal.
  • Step S104 The source signal of the echo signal is used as a reference signal, and the echo estimation signal is generated according to the echo path characteristic parameter.
  • Step S106 subtracting the echo estimation signal from the to-be-processed speech signal.
  • the echo path characteristic parameter of the echo signal can be estimated in real time to generate an echo estimation signal, and the echo estimation signal can be subtracted from the to-be-processed speech signal in real time and adaptively, thereby realizing echo cancellation, and the echo is It may be a downlink echo or an uplink echo; at the same time, in the present invention, the echo cancellation signal is subtracted from the to-be-processed speech signal to eliminate the echo, thereby avoiding the volume of the near-end speaker. Silencing or reducing control to eliminate echoes causes discontinuities in the far-end audio, which helps improve the voice quality of the conference and improves the user experience.
  • the to-be-processed speech signal includes: a near-end speech signal and/or a far-end transmitted speech signal.
  • the to-be-processed voice signal includes: a near-end voice signal and/or a voice signal transmitted from a remote end, that is, the cancellation of the uplink echo can be completed or the downlink echo can be eliminated, for example, when the signal is to be processed.
  • the voice signal is a near-end voice signal
  • the uplink echo is cancelled.
  • the voice signal to be processed is a voice signal transmitted from the far end, the downlink echo is eliminated, so that the uplink echo and/or the downlink echo can be simultaneously performed.
  • the short-term energy estimate and the far-end of the near-end speech signal are acquired within a preset duration before estimating the echo path characteristic parameter of the echo signal.
  • a short-term energy estimate of the transmitted voice signal determining a call state mode according to a short-term energy estimate of the near-end voice signal and a short-term energy estimate of the voice signal transmitted by the far-end, wherein the call state mode Includes near-end and far-end calls.
  • the size determines the call state mode, and the call state mode includes a near-end call and a far-end call, that is, the source signal of the echo signal can be determined after determining the current call state mode, for example, determining that the current call state mode is a near-end call.
  • the source signal of the echo signal is a near-end speech signal; when it is determined that the current call state mode is a far-end call, the source signal of the echo signal is a voice signal transmitted from the far end, and therefore, the call state mode is determined.
  • a call is determined based on the short-term energy estimate of the near-end voice signal and the short-term energy estimate of the voice signal transmitted from the far-end.
  • the state mode method for example, if the short-term energy estimation value of the near-end speech signal is greater than a preset multiple of the short-term energy estimation value of the speech signal transmitted by the far-end within the preset duration, determining the call state mode to be near
  • the voice signal to be processed is the voice signal transmitted by the remote end; otherwise, the call state mode is determined to be the far end call, and the to-be-processed voice signal is the near-end voice signal.
  • the call state mode is determined to be near In the end call, the pending speech signal is a voice signal transmitted from the far end, and the short-term energy estimation value of the near-end speech signal is not greater than a preset multiple of the short-term energy estimation value of the speech signal transmitted from the far end within the preset duration.
  • the call state mode is determined to be a far-end call, and the pending voice signal is a near-end voice signal, thereby accurately determining the call state mode.
  • a method for estimating the echo path characteristic parameter of the echo signal for example, acquiring an echo signal in the to-be-processed speech signal.
  • the echo path characteristic parameter of the echo signal is estimated by considering the delay of the echo signal in the speech signal to be processed and the power error of the speech signal to be processed between the previous time and the current time, for example, in a call.
  • the pending speech signal is the voice signal transmitted from the far end
  • the source signal of the echo signal is the near-end speech signal.
  • the echo path characteristic parameter of the estimated echo signal should consider the source signal of the echo signal.
  • the change of the delay of the echo signal and the change of the signal attenuation, that is, the path characteristic parameter of the process that estimates the source signal of the echo signal is transmitted to the far end through the uplink transmission channel, and is transmitted back to the near end through the far-end echo channel.
  • the characteristic parameters should consider the source signal of the echo signal to become the echo signal.
  • the delay variation of the number, and the change of the signal attenuation, that is, the path characteristic parameter of the source signal of the estimated echo signal from the speaker to the microphone, this process is an estimation of the characteristic parameters of the uplink echo path, thereby accurately estimating the echo path characteristics of the echo signal Parameters, which in turn improve the effect of echo cancellation.
  • a method of considering the delay of the echo signal when determining the characteristic parameters of the echo path is provided, for example, due to uplink echo cancellation and downlink echo cancellation.
  • the difference is that the delay of the echo signal to be processed by the downlink echo cancellation is larger, and accordingly, the number of taps of the adaptive filtering unit is required to be larger, the resource consumption is increased, and the convergence speed is lowered.
  • the echo signal to be processed includes the delay caused by local speech compression coding. This part is a fixed delay, set to ⁇ . When designing the number of filter taps, this part of the delay can be ignored.
  • the fixed delay is directly subtracted before the number of filter taps is designed, so that the tap is no longer designed separately for the fixed delay.
  • Downstream echo cancellation The echo signal to be processed has a delay determined by three parts: 1) near-end processing time; 2) network transmission time; 3) remote processing time. The sum of the three parts of time determines the delay of the echo signal to be processed by the downlink echo cancellation, wherein there are fixed part 1) near-end processing time and 3) far-end processing time, and also variable part 2) network transmission time, thereby
  • the delay time cap and the lower limit design filter for determining the echo signal can be considered as a fixed delay, and the fixed delay is directly subtracted before the number of filter taps is designed, so that the tap is no longer designed separately for the fixed delay.
  • the echo path characteristic parameter is updated if the power error is greater than a preset threshold, wherein the echo path characteristic parameter comprises: an uplink echo path characteristic parameter and/or a downlink echo path characteristic parameter.
  • the echo path is determined to change, for example, the delay of the echo changes. Or the signal attenuation changes.
  • the echo cancellation device includes: an estimation module 202 configured to estimate an echo path characteristic parameter of the echo signal; a generation module 204 connected to the estimation module 202, configured to The source signal of the echo signal is used as a reference signal, and an echo estimation signal is generated according to the echo path characteristic parameter.
  • the processing module 206 is coupled to the generation module 204 and configured to subtract the echo estimation signal from the to-be-processed speech signal.
  • the echo path characteristic parameter of the echo signal is estimated by the estimation module 202.
  • the generating module 204 uses the source signal of the echo signal as a reference signal to generate an echo estimation signal according to the echo path characteristic parameter
  • the processing module 206 is The echo estimation signal is subtracted from the to-be-processed speech signal, so that the echo path characteristic parameters of the echo signal can be estimated in real time to generate an echo estimation signal, and the echo can be subtracted from the to-be-processed speech signal in real time and adaptively.
  • the signal is estimated to achieve the cancellation of the echo.
  • the echo can be either a downlink echo or an uplink echo.
  • the echo estimation signal is subtracted from the to-be-processed speech signal to eliminate the echo.
  • the to-be-processed speech signal includes: a near-end speech signal and/or a far-end transmitted speech signal.
  • a near-end speech signal and/or a far-end transmitted speech signal.
  • the echo cancellation device further includes: an acquisition module 208, configured to acquire a near-end speech signal within a preset duration a short-term energy estimate and a short-term energy estimate of the voice signal transmitted from the far-end; the determining module 210 is coupled to the acquisition module 208, configured to estimate the short-term energy according to the near-end speech signal and the voice transmitted from the far-end
  • the call state mode is determined by the magnitude of the short-term energy estimate of the signal, wherein the call state mode includes a near-end call and a far-end call. In order to accurately determine the call state mode, in the preferred embodiment, as shown in FIG.
  • the determining module 210 includes: a first determining unit 2102, configured to estimate that the short-term energy value of the near-end voice signal is greater than a preset. In the case of a preset multiple of the short-term energy estimation value of the voice signal transmitted from the remote end in the duration, the call state mode is determined to be a near-end call, and the to-be-processed voice signal is a voice signal transmitted from the far-end; the second determining unit 2104 And setting, in a case where the short-term energy estimation value of the near-end speech signal is not greater than a preset multiple of the short-term energy estimation value of the voice signal transmitted from the far-end within the preset duration, determining the call state mode as the far-end call, The speech signal to be processed is a near-end speech signal.
  • the estimation module 202 includes: a first acquiring unit 2022, configured to acquire a to-be-processed voice signal.
  • the second acquisition unit 2024 is configured to obtain a power error between the previous time and the current time of the to-be-processed voice signal;
  • the estimating unit 2026 is connected to the first obtaining unit 2022 and the second acquiring unit 2024. , is set to estimate the echo path characteristic parameters of the echo signal according to the delay and the power error.
  • the preferred embodiment as shown in FIG.
  • the echo cancellation device further includes: an update module 212, configured to determine the to-be-processed speech signal at a previous time and a current time When the power error between the two is greater than the preset threshold, the echo path characteristic parameter is updated, wherein the echo path characteristic parameter comprises: an uplink echo path characteristic parameter and/or a downlink echo path characteristic parameter.
  • an update module 212 configured to determine the to-be-processed speech signal at a previous time and a current time
  • the echo path characteristic parameter is updated, wherein the echo path characteristic parameter comprises: an uplink echo path characteristic parameter and/or a downlink echo path characteristic parameter.
  • the quasi-reconstruction unit 702 is equivalent to the estimation module 202, the generation module 204, the acquisition module 208, the determination module 210, and the update module 212.
  • the subtractor 704 is equivalent to the processing module 206, and the uplink echo and/or can be completed by using the device.
  • the downlink echo is cancelled.
  • the to-be-processed voice signal is a voice signal transmitted from the far end
  • the source signal of the echo signal is a near-end voice signal.
  • the echo path characteristic of the echo signal is estimated.
  • the parameters should consider the change of the source signal of the echo signal into the delay of the echo signal, and the change of the signal attenuation, that is, the source signal of the estimated echo signal is transmitted to the far end through the uplink transmission channel, and then transmitted back to the near end through the far echo channel.
  • the path characteristic parameter of the process which is an estimation of the characteristic parameters of the downlink echo path, and then, using the near-end speech signal as a reference signal, generating an echo estimation signal according to the characteristic parameters of the downlink echo path, and finally, the speech signal transmitted at the far end Subtract the echo estimation signal to complete the downlink echo cancellation
  • the pending speech signal is the near-end speech signal
  • the source signal of the echo signal is the far-end transmitted speech signal.
  • the echo path characteristic parameter of the estimated echo signal should consider the echo.
  • the source signal of the signal changes into the delay of the echo signal, and the change of the signal attenuation, that is, the path characteristic parameter of the source signal of the estimated echo signal from the speaker to the microphone, which is an estimation of the characteristic parameters of the uplink echo path, and then
  • the voice signal transmitted from the far end is a reference signal, and an echo estimation signal is generated according to the characteristic parameters of the uplink echo path.
  • the echo estimation signal is subtracted from the near-end speech signal to complete the elimination of the downlink echo.
  • the echo cancellation device can be designed in accordance with the requirements of the ITU-T G.168 standard.
  • the basic idea of changing the operation of the echo canceling device is to estimate the echo path characteristic parameters of the echo signal to generate a simulated echo estimation signal, and subtract the echo estimation signal from the received signal to be processed, thereby implementing echo cancellation.
  • the signal to be processed containing the echo signal is referred to as S.
  • the source signal that generates the echo signal is referred to as "for the uplink echo cancellation, the near-end speech signal is referred to, Refers to the voice signal transmitted from the far end.
  • For downlink echo cancellation it refers to the voice signal transmitted from the far end, which refers to the near-end voice signal.
  • the above-mentioned echo cancellation device may include the following parts: Voice state detector 802 (equivalent to acquisition module 208 and determination module 210); NLMS (adaptive) controller 804 (equivalent to estimation module 202); adaptive filter 806 (equivalent to generation module 204) and subtractor 808 (equivalent to processing module) 206)
  • Voice state detector 802 Equivalent to acquisition module 208 and determination module 210
  • NLMS adaptive
  • controller 804 equivalent to estimation module 202
  • adaptive filter 806 equivalent to generation module 204
  • subtractor 808 equivalent to processing module 206
  • the above-mentioned voice state detector 802, NLMS (adaptive) controller 804 and adaptive filter 806 are generally equivalent to the echo path parameter estimation and echo simulation reconstruction unit 702 in FIG. 7 (ie, echo path parameter estimation).
  • the echo simulation reconstruction unit does not include the portion in the dotted line frame in Fig. 8), and the above portions complete the echo cancellation by the following communication.
  • the voice state detector 802 determines the current state of the call state of the network according to the signal power of 1 ⁇ and "different time windows.
  • the call state mode includes a near-end call and a far-end call, and the voice state detector 802 will obtain the voice state detector 802.
  • the R i « signal is signaled to the NLMS controller 804, which is based on the previous one.
  • £ delay filter size signal power error signal ( ⁇ echo signal and the time and the current time is estimated echo path characteristic parameters of the echo signal, the adaptive filter 806 and transmitted to the echo path characteristic parameter, adaptive The filter 806 configures the coefficient according to the echo path characteristic parameter, for example, the coefficient is the tap coefficient of the filter, the number of stages, etc., and the source signal of the echo signal is used as a reference signal, and the echo estimation signal is generated according to the configured coefficient. Finally, The generated echo estimation signal is subtracted from the signal to be processed by subtractor 808 to complete the cancellation of the echo.
  • adaptive filter 806 is employed to simulate the echo estimate. The signal can accurately estimate the echo path characteristic parameter and quickly track the change of the echo path characteristic parameter.
  • Step S902 voice state detection
  • the 802 samples the voice signal and the near-end voice signal transmitted from the remote end, and obtains the short-term energy estimation value of the voice signal sample and the near-end voice signal sample transmitted by the remote end according to the sampling, according to the voice transmitted from the remote end.
  • the magnitude of the short-term energy estimate of the signal sample and the near-end speech signal sample determines the call state pattern (information) of the current network, and passes the talk state mode to the NLMS controller 804.
  • the voice state detector 802 can The sampling processing module and the call state judging module are implemented, and the sampling processing module is configured to receive the voice signal and the near-end voice signal transmitted by the remote end, and transmit the voice signal and the near-end voice transmitted to the far end according to the set time period.
  • the signal is sampled to obtain a short-time energy estimate of the voice signal transmitted from the far end and a short-term energy estimate of the near-end voice signal; a call state judgment module, if the short-range voice signal sample point obtained by the sample processing module is short The time energy estimate is greater than the far end transmitted during the set time period
  • the current state of the call state of the network is determined to be the near-end call mode; otherwise, the call state mode of the current network is determined to be the far-end call mode.
  • Step S904 Specifications
  • the normalized Least Mean Squares (NLMS) controller 804 performs coefficients on the sliding window filter (Finite Impulse Response, simply referred to as FIR) according to the current state of the call state mode information transmitted by the voice state detector 802.
  • FIR Finite Impulse Response
  • the coefficient update of the downlink echo cancellation adaptive filter is not performed (same reason, in the case that the current network call state information is the far-end call mode) , then no uplink echo Eliminate the coefficient update of the adaptive filter).
  • the adaptive algorithm employed by the echo cancellation device may be an algorithm group based on the fastest descent method.
  • the representative of this adaptive algorithm is the Least Mean Squares (LMS) algorithm, and the minimum criterion is the root mean square error.
  • LMS Least Mean Squares
  • the NLMS (Energy Normalized Minimum Mean Square Error) algorithm is an improved algorithm of the LMS algorithm, which overcomes the shortcomings of the LMS algorithm being sensitive to input signal energy.
  • the NLMS algorithm and its various improvements are adaptive algorithms in the echo cancellers currently employed.
  • the echo cancellation device described above may employ an NLMS algorithm.
  • Step S906 The adaptive filter 806 uses the source signal of the echo signal as a reference signal according to the coefficient configured by the NLMS controller 804, and performs the voice signal transmitted from the far end filled in the buffer according to the set moving bit length. Step by step filter processing and output.
  • Step S908 The subtracter 808 subtracts the echo estimation signal of the voice signal (source signal of the echo signal) transmitted from the far end from the near-end voice signal (the voice signal to be processed), and outputs the processed signal to other modules. deal with.
  • the voice state detector 802 is required. The determination of the voice state is performed continuously and constantly. Therefore, in addition to the voice state detector 802, the LMS controller 804 and the adaptive filter 806 can be multiplexed, thereby saving digital signal processing (DSP). Resources, whose maximum resource consumption depends on downlink echo cancellation.
  • modules or steps of the present invention can be implemented by a general-purpose computing device, which can be concentrated on a single computing device or distributed over a network composed of multiple computing devices. Alternatively, they may be implemented by program code executable by the computing device, such that they may be stored in the storage device by the computing device and, in some cases, may be different from the order herein.
  • the steps shown or described are performed, or they are separately fabricated into individual integrated circuit modules, or a plurality of modules or steps are fabricated as a single integrated circuit module.
  • the invention is not limited to any specific combination of hardware and software.
  • the above is only the preferred embodiment of the present invention, and is not intended to limit the present invention, and various modifications and changes can be made to the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and scope of the present invention are intended to be included within the scope of the present invention.

Landscapes

  • Engineering & Computer Science (AREA)
  • Signal Processing (AREA)
  • Physics & Mathematics (AREA)
  • Acoustics & Sound (AREA)
  • Multimedia (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Computational Linguistics (AREA)
  • Quality & Reliability (AREA)
  • Health & Medical Sciences (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Human Computer Interaction (AREA)
  • Cable Transmission Systems, Equalization Of Radio And Reduction Of Echo (AREA)
  • Telephone Function (AREA)

Abstract

本发明提供了一种回声消除方法及装置，其中，该方法包括：估计回声信号的回声路径特征参数；以回声信号的来源信号为参考信号，按照回声路径特征参数生成回声估计信号；在待处理语音信号中减去回声估计信号。本发明解决了相关技术中在抑制回传的近端音频时无法做到自适应的问题，从而有助于提高会议的语音质量，改善用户体验。

Description

回声消除方法及装置 技术领域 本发明涉及通信领域， 具体而言， 涉及一种回声消除方法及装置。 背景技术 音频质量一直是困扰电话会议的一个大问题。 在双方通话过程中， 会议终端之间 一般由两个通道构成，其中， 下行接收通道将对端传送过来的语音信号通过语音解码、 重构、 放大后交给扬声器播放出来， 上行发送通道将近端用户的语音信号通过麦克风 采集进来， 经过语音处理、 编码后发送给会议的对端。 在实际工作当中， 由于会议双 方有同时说话的需要， 因此， 扬声器和麦克风可能同时都在工作， 如图 1所示。 在会 议双方同时说话的场景下， 由于会议终端尺寸的限制， 扬声器播放的远端语音信号有 可能被近端麦克风采集， 并通过上行发送通道传送给远端通话者， 由于通过上行发送 通道将远端语音信号再传回远端存在一定的时延， 此时， 远端通话者就有可能听到自 己的讲话， 即回音， 并且时延越大， 回音越明显。 因此回音对 VOIP这类语音通话影 响最大。 为了克服回音对语音通话的影响， 目前有许多种回音消除技术， 例如专利号为
CN200610114419.5， CN200710100270.X, CN200820213203.9， CN200880104273.3, CN201010225201.3 ， CN201010235614.X, CN201010240571.4 ， CN201110048861.3 等专利申请文件都对回音消除技术进行了公开， 但是， 上述专利申请文件中公开的回 音消除技术的共同点在于近端参与者为远端用户做了回声消除， 有助于防止远端的参 与者听到他自己的声音作为回音回传到远端， 尽管在近端单元处可能使用了近端回音 消除， 使得远端的参与者不会听到他自己的声音， 然而， 远端单元在一些情况下可能 没有可用的声学回音消除器来消除远端回音， 在这种情况下， 由于在远端的扬声器和 麦克风之间的声学耦合， 近端参与者将会听到他自己的声音回传到近端。 因此， 近端 回音消除可能使远端的参与者受益， 但对防止近端的参与者听到作为回音从远端被回 传的近端音频却没有用。 专利号为 CN201010287069.9 的专利申请文件提出了一种在近端对远端回传的近 端干扰进行抑制的方法。该专利申请文件公开了涉及在近端对回传音频的检测和抑制， 即来自近端、 在远端被声学耦合并被回传到近端单元的音频在会议的近端被检测和抑 制。 为了从近端单元发送的近端音频和在近端单元接收的远端音频中确定出各分离频 带的第一能量输出和第二能量输出， 近端单元在某个时间延迟范围内对每个频带的第 一能量输出和第二能量输出进行相互比较， 并基于该比较结果来检测在接收到的远端 音频中被发送的近端音频的回传。 该比较可使用互相关来得到估计的时间延迟， 以用 于近端能量和远端能量的进一步分析。 近端单元通过消音或减弱在其扬声器处输出的 远端音频来抑制任何被检测到的回传的近端音频。 即其核心思想是： 通过双端讲话检 测器单元， 如果在近端没有检测到远端参与者的语音， 则近端扬声器可被消音或降低 音量以抑制回传的近端音频。 这种方法的确可以有效地抑制回传的近端音频， 但是该 方法并不能自适应地来抑制回传的近端音频， 不能实时地对近端扬声器的音量进行控 制； 此外， 该抑制回传的近端音频的方法是对近端扬声器的音量进行消音或降低控制， 从而可能导致远端音频的不连续， 进而降低会议的语音质量， 影响用户体验。 因此， 在上述相关技术中， 在抑制回传的近端音频时存在由于无法自适应地对近 端扬声器的音量进行消音或降低控制而导致远端音频的不连续， 进而降低会议的语音 质量， 影响用户体验的问题。 发明内容 本发明提供了一种回声消除方法及装置， 以至少解决相关技术中在抑制回传的近 端音频时无法做到自适应的问题。 根据本发明的一个方面， 提供了一种回声消除方法， 其包括： 估计回声信号的回 声路径特征参数； 以回声信号的来源信号为参考信号， 按照回声路径特征参数生成回 声估计信号； 在待处理语音信号中减去回声估计信号。 优选地， 待处理语音信号包括： 近端语音信号和 /或远端传输过来的语音信号。 优选地， 在估计回声信号的回声路径特征参数之前， 上述回声消除方法还包括： 在预设时长内获取近端语音信号的短时能量估计值和远端传输过来的语音信号的短时 能量估计值； 根据近端语音信号的短时能量估计值和远端传输过来的语音信号的短时 能量估计值的大小来确定通话状态模式， 其中， 通话状态模式包括近端通话和远端通 话。 优选地， 根据近端语音信号的短时能量估计值和远端传输过来的语音信号的短时 能量估计值的大小来确定通话状态模式包括： 在近端语音信号的短时能量估计值大于 预设时长内远端传输过来的语音信号的短时能量估计值的预设倍数的情况下， 确定通 话状态模式为近端通话， 待处理语音信号为远端传输过来的语音信号； 否则， 确定通 话状态模式为远端通话， 待处理语音信号为近端语音信号。 优选地， 估计回声信号的回声路径特征参数包括： 获取待处理语音信号中回声信 号的时延； 获取待处理语音信号在前一时刻和当前时刻之间的功率误差； 按照时延和 功率误差来估计回声信号的回声路径特征参数。 优选地， 在按照时延和功率误差来估计回声信号的回声路径特征参数之后， 上述 回声消除方法还包括： 确定待处理语音信号在前一时刻和当前时刻之间的功率误差大 于预设阈值的情况下， 更新回声路径特征参数， 其中， 回声路径特征参数包括： 上行 回声路径特征参数和 /或下行回声路径特征参数。 根据本发明的另一方面， 提供了一种回声消除装置， 其包括： 估计模块， 设置为 估计回声信号的回声路径特征参数； 生成模块， 设置为以回声信号的来源信号为参考 信号， 按照回声路径特征参数生成回声估计信号； 处理模块， 设置为在待处理语音信 号中减去回声估计信号。 优选地， 上述回声消除装置还包括： 获取模块， 设置为在预设时长内获取近端语 音信号的短时能量估计值和远端传输过来的语音信号的短时能量估计值； 确定模块， 设置为根据近端语音信号的短时能量估计值和远端传输过来的语音信号的短时能量估 计值的大小来确定通话状态模式， 其中， 通话状态模式包括近端通话和远端通话。 优选地， 确定模块包括： 第一确定单元， 设置为在近端语音信号的短时能量估计 值大于预设时长内远端传输过来的语音信号的短时能量估计值的预设倍数的情况下， 确定通话状态模式为近端通话， 待处理语音信号为远端传输过来的语音信号； 第二确 定单元， 设置为在近端语音信号的短时能量估计值不大于预设时长内远端传输过来的 语音信号的短时能量估计值的预设倍数的情况下， 确定通话状态模式为远端通话， 待 处理语音信号为近端语音信号。 优选地， 估计模块包括： 第一获取单元， 设置为获取待处理语音信号中回声信号 的时延； 第二获取单元， 设置为获取待处理语音信号在前一时刻和当前时刻之间的功 率误差； 估计单元， 设置为按照时延和功率误差来估计回声信号的回声路径特征参数。 优选地， 上述回声消除装置还包括： 更新模块， 设置为在确定待处理语音信号在 前一时刻和当前时刻之间的功率误差大于预设阈值的情况下，更新回声路径特征参数， 其中， 回声路径特征参数包括： 上行回声路径特征参数和 /或下行回声路径特征参数。 在本发明中， 通过估计回声信号的回声路径特征参数， 然后， 以回声信号的来源 信号为参考信号， 按照回声路径特征参数生成回声估计信号， 最终， 在待处理语音信 号中减去回声估计信号，实现了可以实时地对回声信号的回声路径特征参数进行估计， 以便生成回声估计信号， 进而可以实时地、 自适应地在待处理语音信号中减去回声估 计信号， 实现回声的消除， 上述回声既可以是下行回声也可以是上行回声； 同时， 在 本发明中， 是通过在待处理语音信号中减去回声估计信号来达到消除回声的目的的， 因此， 避免了通过对近端扬声器的音量进行消音或降低控制来消除回声而导致远端音 频的不连续的问题， 从而有助于提高会议的语音质量， 改善用户体验。 附图说明 此处所说明的附图用来提供对本发明的进一步理解， 构成本申请的一部分， 本发 明的示意性实施例及其说明用于解释本发明， 并不构成对本发明的不当限定。 在附图 中- 图 1是根据本发明实施例的回声消除方法的流程图； 图 2是根据本发明实施例的回声消除装置的结构框图； 图 3是根据本发明实施例的另一种回声消除装置的结构框图； 图 4是根据本发明实施例的确定模块的结构框图； 图 5是根据本发明实施例的估计模块的结构框图； 图 6是根据本发明实施例的又一种回声消除装置的结构框图； 图 7是根据本发明实施例的回声消除系统的功能示意图； 图 8是根据本发明实施例的回声消除装置的功能框图； 以及 图 9是根据本发明实施例的上行回声消除方法的流程图。 具体实施方式 下文中将参考附图并结合实施例来详细说明本发明。 需要说明的是， 在不冲突的 情况下， 本申请中的实施例及实施例中的特征可以相互组合。 本实施例提供了一种回声消除方法， 如图 1所示， 该回声消除方法包括步骤 S102 至步骤 S106。 步骤 S102: 估计回声信号的回声路径特征参数。 步骤 S104: 以回声信号的来源信号为参考信号， 按照回声路径特征参数生成回声 估计信号。 步骤 S106: 在待处理语音信号中减去回声估计信号。 通过上述步骤， 首先估计回声信号的回声路径特征参数， 然后， 以回声信号的来 源信号为参考信号， 按照回声路径特征参数生成回声估计信号， 最终， 在待处理语音 信号中减去回声估计信号， 实现了可以实时地对回声信号的回声路径特征参数进行估 计， 以便生成回声估计信号， 进而可以实时地、 自适应地在待处理语音信号中减去回 声估计信号， 实现回声的消除， 上述回声既可以是下行回声也可以是上行回声； 同时， 在本发明中，是通过在待处理语音信号中减去回声估计信号来达到消除回声的目的的， 因此， 避免了通过对近端扬声器的音量进行消音或降低控制来消除回声而导致远端音 频的不连续的问题， 从而有助于提高会议的语音质量， 改善用户体验。 为了提高本实施例的实用性， 在本优选实施例中， 上述待处理语音信号包括： 近 端语音信号和 /或远端传输过来的语音信号。 在上述优选实施例中， 上述待处理语音信号包括： 近端语音信号和 /或远端传输过 来的语音信号， 即既可以完成上行回声的消除也可以完成下行回声的消除， 例如， 当 待处理语音信号为近端语音信号时， 则完成上行回声的消除； 当待处理语音信号为远 端传输过来的语音信号时， 则完成下行回声的消除， 从而可以同时进行上行回声和 / 或下行回声的消除， 有效地提高会议的语音质量， 提高了本实施例的实用性。 为了准确地估计出回声信号的回声路径特征参数， 在本优选实施例中， 在估计回 声信号的回声路径特征参数之前， 在预设时长内获取近端语音信号的短时能量估计值 和远端传输过来的语音信号的短时能量估计值； 根据近端语音信号的短时能量估计值 和远端传输过来的语音信号的短时能量估计值的大小来确定通话状态模式， 其中， 通 话状态模式包括近端通话和远端通话。 在上述优选实施例中， 在估计回声信号的回声路径特征参数之前， 通过根据在预 设时长内近端语音信号的短时能量估计值和远端传输过来的语音信号的短时能量估计 值的大小来确定通话状态模式， 该通话状态模式包括近端通话和远端通话， 即确定当 前的通话状态模式后即可确定回声信号的来源信号， 例如， 在确定当前的通话状态模 式为近端通话时， 则回声信号的来源信号为近端语音信号； 在确定当前的通话状态模 式为远端通话时， 则回声信号的来源信号为远端传输过来的语音信号， 因此， 在确定 了通话状态模式时， 进而确定了回声信号的来源信号， 从而可以准确地估计出回声信 号的回声路径特征参数。 为了准确地确定出通话状态模式， 在本优选实施例中， 提供了一种根据近端语音 信号的短时能量估计值和远端传输过来的语音信号的短时能量估计值的大小来确定通 话状态模式的方法， 例如， 在近端语音信号的短时能量估计值大于预设时长内远端传 输过来的语音信号的短时能量估计值的预设倍数的情况下， 确定通话状态模式为近端 通话， 待处理语音信号为远端传输过来的语音信号； 否则， 确定通话状态模式为远端 通话， 待处理语音信号为近端语音信号。 在上述优选实施例中， 在近端语音信号的短时能量估计值大于预设时长内远端传 输过来的语音信号的短时能量估计值的预设倍数的情况下， 确定通话状态模式为近端 通话， 待处理语音信号为远端传输过来的语音信号， 在近端语音信号的短时能量估计 值不大于预设时长内远端传输过来的语音信号的短时能量估计值的预设倍数的情况 下， 确定通话状态模式为远端通话， 待处理语音信号为近端语音信号， 从而准确地确 定出通话状态模式。 为了提高回声消除的效果， 准确地估计回声信号的回声路径特征参数， 在本优选 实施例中， 提供了一种估计回声信号的回声路径特征参数的方法， 例如， 获取待处理 语音信号中回声信号的时延； 获取待处理语音信号在前一时刻和当前时刻之间的功率 误差； 按照时延和功率误差来估计回声信号的回声路径特征参数。 在上述优选实施例中， 同时考虑待处理语音信号中回声信号的时延和待处理语音 信号在前一时刻和当前时刻之间的功率误差来估计回声信号的回声路径特征参数， 例 如， 在通话状态模式为近端通话时， 待处理语音信号为远端传输过来的语音信号， 回 声信号的来源信号为近端语音信号， 此时估计回声信号的回声路径特征参数则要考虑 回声信号的来源信号变为回声信号的时延变化和信号衰减变化， 即估计回声信号的来 源信号通过上行发送通道传送至远端， 通过远端回声通道又回传到近端的过程的路径 特征参数， 此过程为下行回声路径特征参数的估计； 在通话状态模式为远端通话时， 待处理语音信号为近端语音信号， 回声信号的来源信号为远端传输过来的语音信号， 此时估计回声信号的回声路径特征参数则要考虑回声信号的来源信号变为回声信号的 时延变化， 和信号衰减变化， 即估计回声信号的来源信号由扬声器到麦克风之间的路 径特征参数， 此过程为上行回声路径特征参数的估计， 从而准确地估计回声信号的回 声路径特征参数， 进而提高回声消除的效果。 优选地， 为了降低计算量， 缩短系统响应时间， 在本优选实施例中， 提供了一种 在确定回声路径特征参数时考虑回声信号的时延的方法， 例如， 由于上行回声消除和 下行回声消除从本质上来看是一样的， 区别在于下行回声消除要处理的回声信号时延 更大， 相应地要求自适应滤波单元的抽头数更多， 增加资源消耗， 降低收敛速度。 上 行回声消除要处理的回声信号其时延包括本地语音压缩编码造成的时延， 这部分是 个固定时延， 设为 ^， 在设计滤波器抽头数时可以不考虑此部分时延， 而是在设计滤 波器抽头数前直接将该固定时延减去， 从而不再单独为该固定时延设计抽头。 下行回 声消除要处理的回声信号其时延由三部分共同决定： 1 ) 近端处理时间； 2) 网络传输 时间； 3 )远端处理时间。三部分时间相加共同决定了下行回声消除要处理的回声信号 的时延， 其中有固定部分 1 )近端处理时间和 3 )远端处理时间， 也有可变部分 2) 网 络传输时间，由此可确定回声信号的时延上限 皿和下限 设计滤波器时可将 考虑为固定时延， 在设计滤波器抽头数前直接将该固定时延减去， 从而不再单独为该 固定时延设计抽头， 通过这一措施可以极大的降低计算量， 缩短系统响应时间。 为了实时地跟踪回声路径特征参数的变化， 在本优选实施例中， 在按照时延和功 率误差来估计回声信号的回声路径特征参数之后， 确定待处理语音信号在前一时刻和 当前时刻之间的功率误差大于预设阈值的情况下， 更新回声路径特征参数， 其中， 回 声路径特征参数包括： 上行回声路径特征参数和 /或下行回声路径特征参数。 在上述优选实施例中， 当确定待处理语音信号在前一时刻和当前时刻之间的功率 误差大于预设阈值的情况下， 即确定回声路径发生了变化， 例如， 回声的时延发生变 化， 或信号衰减发生变化， 此时需要实时地更新回声路径特征参数， 以便在准确地估 计出回声路径特征参数后可以实时地跟踪回声路径特征参数的变化， 进而确保有效地 消除回声。 本实施例提供了一种回声消除装置， 如图 2所示， 该回声消除装置包括： 估计模 块 202， 设置为估计回声信号的回声路径特征参数； 生成模块 204， 连接至估计模块 202， 设置为以回声信号的来源信号为参考信号， 按照回声路径特征参数生成回声估计 信号； 处理模块 206， 连接至生成模块 204， 设置为在待处理语音信号中减去回声估计 信号。 在上述实施例中， 通过估计模块 202估计回声信号的回声路径特征参数， 然后， 生成模块 204以回声信号的来源信号为参考信号， 按照回声路径特征参数生成回声估 计信号， 最终， 处理模块 206在待处理语音信号中减去回声估计信号， 实现了可以实 时地对回声信号的回声路径特征参数进行估计， 以便生成回声估计信号， 进而可以实 时地、 自适应地在待处理语音信号中减去回声估计信号， 实现回声的消除， 上述回声 既可以是下行回声也可以是上行回声； 同时， 在本发明中， 是通过在待处理语音信号 中减去回声估计信号来达到消除回声的目的的， 因此， 避免了通过对近端扬声器的音 量进行消音或降低控制来消除回声而导致远端音频的不连续的问题， 从而有助于提高 会议的语音质量， 改善用户体验。 为了提高本实施例的实用性， 在本优选实施例中， 上述待处理语音信号包括： 近 端语音信号和 /或远端传输过来的语音信号。 为了准确地估计出回声信号的回声路径特征参数， 在本优选实施例中， 如图 3所 示， 上述回声消除装置还包括： 获取模块 208， 设置为在预设时长内获取近端语音信 号的短时能量估计值和远端传输过来的语音信号的短时能量估计值； 确定模块 210， 连接至获取模块 208， 设置为根据近端语音信号的短时能量估计值和远端传输过来的 语音信号的短时能量估计值的大小来确定通话状态模式， 其中， 通话状态模式包括近 端通话和远端通话。 为了准确地确定出通话状态模式， 在本优选实施例中， 如图 4所示， 上述确定模 块 210包括： 第一确定单元 2102， 设置为在近端语音信号的短时能量估计值大于预设 时长内远端传输过来的语音信号的短时能量估计值的预设倍数的情况下， 确定通话状 态模式为近端通话，待处理语音信号为远端传输过来的语音信号；第二确定单元 2104， 设置为在近端语音信号的短时能量估计值不大于预设时长内远端传输过来的语音信号 的短时能量估计值的预设倍数的情况下， 确定通话状态模式为远端通话， 待处理语音 信号为近端语音信号。 为了提高回声消除的效果， 准确地估计回声信号的回声路径特征参数， 在本优选 实施例中， 如图 5所示， 上述估计模块 202包括： 第一获取单元 2022， 设置为获取待 处理语音信号中回声信号的时延； 第二获取单元 2024， 设置为获取待处理语音信号在 前一时刻和当前时刻之间的功率误差； 估计单元 2026， 连接至第一获取单元 2022和 第二获取单元 2024， 设置为按照时延和功率误差来估计回声信号的回声路径特征参 数。 为了实时地跟踪回声路径特征参数的变化， 在本优选实施例中， 如图 6所示， 上 述回声消除装置还包括： 更新模块 212， 设置为在确定待处理语音信号在前一时刻和 当前时刻之间的功率误差大于预设阈值的情况下， 更新回声路径特征参数， 其中， 回 声路径特征参数包括： 上行回声路径特征参数和 /或下行回声路径特征参数。 以下结合附图和实例对上述各优选实施例进行详细地描述。 如图 7所示， 上述回声消除装置在完成回声消除的过程中可以通过回声路径参数 估计及回声模拟重构单元 702和一个减法器 704来实现， 回声路径参数估计及回声模 拟重构单元 702相当于上述的估计模块 202、 生成模块 204、 获取模块 208、 确定模块 210以及更新模块 212， 减法器 704相当于处理模块 206， 通过该装置的使用可以完成 上行回声和 /或下行回声的消除， 例如， 在通话状态模式为近端通话时， 待处理语音信 号为远端传输过来的语音信号， 回声信号的来源信号为近端语音信号， 此时估计回声 信号的回声路径特征参数则要考虑回声信号的来源信号变为回声信号的时延变化， 和 信号衰减变化， 即估计回声信号的来源信号通过上行发送通道传送至远端， 通过远端 回声通道又回传到近端的过程的路径特征参数， 此过程为下行回声路径特征参数的估 计， 然后， 以近端语音信号为参考信号， 按照下行回声路径特征参数生成回声估计信 号， 最后， 在远端传输过来的语音信号中减去回声估计信号， 以完成下行回声的消除; 在通话状态模式为远端通话时， 待处理语音信号为近端语音信号， 回声信号的来源信 号为远端传输过来的语音信号， 此时估计回声信号的回声路径特征参数则要考虑回声 信号的来源信号变为回声信号的时延变化， 和信号衰减变化， 即估计回声信号的来源 信号由扬声器到麦克风之间的路径特征参数，此过程为上行回声路径特征参数的估计， 然后， 以远端传输过来的语音信号为参考信号， 按照上行回声路径特征参数生成回声 估计信号， 最后， 在近端语音信号中减去回声估计信号， 以完成下行回声的消除。 在本实例中回声消除装置可以是根据 ITU-T G.168标准要求设计的。 改回声消除 装置工作的基本思想是估计回声信号的回声路径特征参数， 以产生一个模拟的回声估 计信号， 从接收到的待处理信号中减去该回声估计信号， 从而实现回声消除。 为了统 一叙述上行回声和下行回声的方便， 在本实例中将含有回声信号的待处理信号称为 S 将产生回声信号的来源信号称为 "， 对于上行回声消除而言， 指近端语音信 号， 指远端传送过来的语音信号， 对于下行回声消除而言， 指远端传送过来的 语音信号， 指近端语音信号。 如图 8所示， 上述回声消除装置可以包括如下部分： 语音状态检测器 802 (相当于获取模块 208和确定模块 210); NLMS (自适应）控制 器 804 (相当于估计模块 202); 自适应滤波器 806 (相当于生成模块 204) 以及减法器 808 (相当于处理模块 206)。 上述语音状态检测器 802、 NLMS (自适应） 控制器 804 和自适应滤波器 806三者整体相当于图 7中的回声路径参数估计及回声模拟重构单元 702 (即回声路径参数估计及回声模拟重构单元不包括图 8中虚线框内的部分）， 上述 各部分通过以下通信来完成回声的消除。 首先， 语音状态检测器 802根据1 ^和 "不同时间窗的信号功率大小来判断当前 网络的通话状态模式， 该通话状态模式包括近端通话和远端通话， 同时语音状态检测 器 802将获得的 和 Ri«信号通知给 NLMS控制器 804， NLMS控制器 804根据前一 时刻和当前时刻滤波器中的 信号的功率误差信号的大小 £ (^和回声信号的时延来 估计回声信号的回声路径特征参数， 并将回声路径特征参数传送给自适应滤波器 806， 自适应滤波器 806根据回声路径特征参数来配置系数， 例如， 该系数为滤波器的抽头 系数， 级数等， 并以回声信号的来源信号 "为参考信号， 按照配置的系数生成回声估 计信号， 最后， 通过减法器 808在待处理信号 "中减去生成的回声估计信号， 以完成 回声的消除。 优选地， 由于回声路径通常是时变的和未知的， 所以采用自适应滤波器 806来模 拟回声估计信号可以准确地估计出回声路径特征参数的同时， 迅速的跟踪回声路径特 征参数的变化。 在本实例中以上行回声的消除为例来详细地描述上述回声消除方法的， 如图 9所 示， 该回声消除方法包括步骤 S902至步骤 S908。 步骤 S902:语音状态检测器 802对远端传输过来的语音信号和近端语音信号进行 采样， 并根据采样获得远端传输过来的语音信号样本和近端语音信号样本的短时能量 估计值， 根据远端传输过来的语音信号样本和近端语音信号样本的短时能量估计值的 大小确定当前网络的通话状态模式 （信息）， 并将该通话状态模式传递给 NLMS控制 器 804。 优选地，上述语音状态检测器 802可以由采样处理模块和通话状态判断模块实现， 采样处理模块， 设置为接收远端传输过来的语音信号和近端语音信号， 并根据设 定的时间段对远端传输过来的语音信号和近端语音信号进行采样处理， 获得远端传输 过来的语音信号的短时能量估计值和近端语音信号的短时能量估计值； 通话状态判断模块， 如果采样处理模块获得的近端语音信号样本点的短时能量估 计值大于在上述设定时间段内的远端传输过来的语音信号的短时能量估计值中最大值 的设定倍数时， 则确定当前网络的通话状态模式为近端通话模式； 否则， 确定当前网 络的通话状态模式为远端通话模式。 步骤 S904: 规格化最小均方差 （Normalized Least Mean Squares, 简称为 NLMS) 控制器 804根据语音状态检测器 802传递过来的当前网络的通话状态模式信息， 对滑 动窗滤波器 （Finite Impulse Response, 简称为 FIR) 进行系数配置； 在当前网络的通 话状态信息为近端通话模式的情况下， 则不进行下行回声消除自适应滤波器的系数更 新 （同理， 在当前网络的通话状态信息为远端通话模式的情况下， 则不进行上行回声 消除自适应滤波器的系数更新）。回声消除装置采用的自适应算法可以为基于最快下降 法的算法群。这种自适应算法的代表为最小均方差（Least Mean Squares,简称为 LMS) 算法， 其最小化准则为均方根误差。 NLMS (能量归一化最小均方误差） 算法是 LMS 算法的改进算法，它克服了 LMS算法对输入信号能量敏感的缺点。 NLMS算法和它的 各种改进形式， 为目前主要采用的回声消除器中的自适应算法。 优选地， 上述回声消 除装置可以采用 NLMS算法。 步骤 S906: 自适应滤波器 806根据 NLMS控制器 804配置的系数， 以回声信号的 来源信号为参考信号， 按照设定的移动位长， 对缓冲区中填入的远端传输过来的语音 信号进行逐步滤波处理后输出。 步骤 S908: 减法器 808从近端语音信号 （待处理语音信号） 中减去远端传输过来 的语音信号 （回声信号的来源信号） 的回声估计信号， 并将处理后的信号输出给其它 模块进行处理。 优选地， 在上述回声消除装置中， 由于上行回声消除和下行回声消除在本质上来 看是一致的， 而且上行回声消除和下行回声消除可以设计成分时工作的， 因此， 由于 语音状态检测器 802需要不停地、 时时地进行语音状态的确定， 所以， 除语音状态检 测器 802外， LMS控制器 804， 自适应滤波器 806可以复用， 从而节约数字信号处理 (Digital Signal Processing,简称为 DSP)资源，其最大资源消耗取决于下行回声消除。 显然， 本领域的技术人员应该明白， 上述的本发明的各模块或各步骤可以用通用 的计算装置来实现， 它们可以集中在单个的计算装置上， 或者分布在多个计算装置所 组成的网络上， 可选地， 它们可以用计算装置可执行的程序代码来实现， 从而， 可以 将它们存储在存储装置中由计算装置来执行， 并且在某些情况下， 可以以不同于此处 的顺序执行所示出或描述的步骤， 或者将它们分别制作成各个集成电路模块， 或者将 它们中的多个模块或步骤制作成单个集成电路模块来实现。 这样， 本发明不限制于任 何特定的硬件和软件结合。 以上所述仅为本发明的优选实施例而已， 并不用于限制本发明， 对于本领域的技 术人员来说， 本发明可以有各种更改和变化。 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， 均应包含在本发明的保护范围之内。

Claims

权 利 要 求 书
1. 一种回声消除方法， 包括：
估计回声信号的回声路径特征参数；
以所述回声信号的来源信号为参考信号， 按照所述回声路径特征参数生成 回声估计信号；
在待处理语音信号中减去所述回声估计信号。
2. 根据权利要求 1所述的方法， 其中， 所述待处理语音信号包括： 近端语音信号 和 /或远端传输过来的语音信号。
3. 根据权利要求 1或 2所述的方法， 其中， 在估计所述回声信号的回声路径特征 参数之前， 还包括：
在预设时长内获取近端语音信号的短时能量估计值和远端传输过来的语音 信号的短时能量估计值；
根据所述近端语音信号的短时能量估计值和所述远端传输过来的语音信号 的短时能量估计值的大小来确定通话状态模式， 其中， 所述通话状态模式包括 近端通话和远端通话。
4. 根据权利要求 3所述的方法， 其中， 所述根据所述近端语音信号的短时能量估 计值和所述远端传输过来的语音信号的短时能量估计值的大小来确定通话状态 模式包括- 在所述近端语音信号的短时能量估计值大于所述预设时长内所述远端传输 过来的语音信号的短时能量估计值的预设倍数的情况下， 确定通话状态模式为 近端通话， 所述待处理语音信号为远端传输过来的语音信号；
否则， 确定通话状态模式为远端通话， 所述待处理语音信号为近端语音信 号。
5. 根据权利要求 1所述的方法， 其中， 所述估计所述回声信号的回声路径特征参 数包括：
获取所述待处理语音信号中回声信号的时延；
获取所述待处理语音信号在前一时刻和当前时刻之间的功率误差； 按照所述时延和所述功率误差来估计所述回声信号的回声路径特征参数。 根据权利要求 5所述的方法， 其中， 在按照所述时延和所述功率误差来估计所 述回声信号的回声路径特征参数之后， 还包括- 确定所述待处理语音信号在前一时刻和当前时刻之间的功率误差大于预设 阈值的情况下， 更新所述回声路径特征参数， 其中， 所述回声路径特征参数包 括： 上行回声路径特征参数和 /或下行回声路径特征参数。 一种回声消除装置， 包括：
估计模块， 设置为估计回声信号的回声路径特征参数；
生成模块， 设置为以所述回声信号的来源信号为参考信号， 按照所述回声 路径特征参数生成回声估计信号；
处理模块， 设置为在待处理语音信号中减去所述回声估计信号。 根据权利要求 7所述的装置， 还包括：
获取模块， 设置为在预设时长内获取近端语音信号的短时能量估计值和远 端传输过来的语音信号的短时能量估计值；
确定模块， 设置为根据所述近端语音信号的短时能量估计值和所述远端传 输过来的语音信号的短时能量估计值的大小来确定通话状态模式， 其中， 所述 通话状态模式包括近端通话和远端通话。 根据权利要求 8所述的装置， 其中， 所述确定模块包括：
第一确定单元， 设置为在所述近端语音信号的短时能量估计值大于所述预 设时长内所述远端传输过来的语音信号的短时能量估计值的预设倍数的情况 下， 确定通话状态模式为近端通话， 所述待处理语音信号为远端传输过来的语 音信号；
第二确定单元， 设置为在所述近端语音信号的短时能量估计值不大于所述 预设时长内所述远端传输过来的语音信号的短时能量估计值的预设倍数的情况 下， 确定通话状态模式为远端通话， 所述待处理语音信号为近端语音信号。 根据权利要求 7所述的装置， 其中， 所述估计模块包括：
第一获取单元， 设置为获取所述待处理语音信号中回声信号的时延； 第二获取单元， 设置为获取所述待处理语音信号在前一时刻和当前时刻之 间的功率误差；
估计单元， 设置为按照所述时延和所述功率误差来估计所述回声信号的回 声路径特征参数。
11. 根据权利要求 10所述的装置， 还包括：
更新模块， 设置为在确定所述待处理语音信号在前一时刻和当前时刻之间 的功率误差大于预设阈值的情况下， 更新所述回声路径特征参数， 其中， 所述 回声路径特征参数包括： 上行回声路径特征参数和 /或下行回声路径特征参数。
PCT/CN2012/077022 2012-05-10 2012-06-15 回声消除方法及装置 WO2013166761A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
EP12876306.7A EP2849351B1 (en) 2012-05-10 2012-06-15 Echo elimination method and device
US14/400,289 US9443528B2 (en) 2012-05-10 2012-06-15 Method and device for eliminating echoes

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201210145202.6A CN103391381B (zh) 2012-05-10 2012-05-10 回声消除方法及装置
CN201210145202.6 2012-05-10

Publications (1)

Publication Number Publication Date
WO2013166761A1 true WO2013166761A1 (zh) 2013-11-14

Family

ID=49535534

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2012/077022 WO2013166761A1 (zh) 2012-05-10 2012-06-15 回声消除方法及装置

Country Status (4)

Country Link
US (1) US9443528B2 (zh)
EP (1) EP2849351B1 (zh)
CN (1) CN103391381B (zh)
WO (1) WO2013166761A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104778950A (zh) * 2014-01-15 2015-07-15 华平信息技术股份有限公司 一种基于回声消除的麦克风信号延时补偿控制方法
CN110956975A (zh) * 2019-12-06 2020-04-03 展讯通信（上海）有限公司 回声消除方法及装置

Families Citing this family (31)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE112013007077T5 (de) * 2013-05-14 2016-02-11 Mitsubishi Electric Corporation Echoauslöschungsvorrichtung
US8953777B1 (en) * 2013-05-30 2015-02-10 Amazon Technologies, Inc. Echo path change detector with robustness to double talk
CN106165015B (zh) * 2014-01-17 2020-03-20 英特尔公司 用于促进基于加水印的回声管理的装置和方法
GB2519392B (en) * 2014-04-02 2016-02-24 Imagination Tech Ltd Auto-tuning of an acoustic echo canceller
GB2521881B (en) 2014-04-02 2016-02-10 Imagination Tech Ltd Auto-tuning of non-linear processor threshold
CN105227900A (zh) * 2014-06-25 2016-01-06 中兴通讯股份有限公司 音视频监控装置、系统及基于音视频监控装置的通话方法
CN104167212A (zh) * 2014-08-13 2014-11-26 深圳市泛海三江科技发展有限公司 一种智能楼宇系统的音频处理方法及装置
FR3025923A1 (fr) * 2014-09-12 2016-03-18 Orange Discrimination et attenuation de pre-echos dans un signal audionumerique
CN105530390B (zh) * 2014-09-30 2018-07-31 华为技术有限公司 会议服务器及其检测会议中的回声来源的方法
CN105991857A (zh) * 2015-02-12 2016-10-05 中兴通讯股份有限公司 一种实现参考信号调整的方法及装置
CN106297816B (zh) * 2015-05-20 2019-12-13 广州质音通讯技术有限公司 一种回声消除的非线性处理方法和装置及电子设备
CN106470284B (zh) * 2015-08-20 2020-02-11 钉钉控股（开曼）有限公司 消除声学回声的方法、装置、系统、服务器及通话装置
CN105915738A (zh) * 2016-05-30 2016-08-31 宇龙计算机通信科技(深圳)有限公司 回声消除方法、回声消除装置和终端
CN106210369A (zh) * 2016-07-15 2016-12-07 福州米立科技有限公司 应用于楼宇对讲系统的消除回声方法及装置
CN106791244B (zh) * 2016-12-13 2020-03-27 青岛微众在线网络科技有限公司 回声消除方法、装置以及通话设备
CN108831491B (zh) * 2017-05-04 2020-07-17 展讯通信（上海）有限公司 回声延迟估计方法及装置、存储介质、电子设备
CN107819964B (zh) * 2017-11-10 2021-04-06 Oppo广东移动通信有限公司 提高通话质量的方法、装置、终端和计算机可读存储介质
CN109961797B (zh) * 2017-12-25 2023-07-18 阿里巴巴集团控股有限公司 一种回声消除方法、装置以及电子设备
CN108055417B (zh) * 2017-12-26 2020-09-29 杭州叙简科技股份有限公司 一种基于语音检测回音抑制切换音频处理系统及方法
US10708689B2 (en) * 2018-05-15 2020-07-07 LogMeln, Inc. Reducing acoustic feedback over variable-delay pathway
CN110913312B (zh) * 2018-09-17 2021-06-18 海信集团有限公司 一种回声消除方法及装置
CN109545237B (zh) * 2018-10-24 2022-01-28 广东思派康电子科技有限公司 一种计算机可读存储介质和应用该介质的语音交互音箱
WO2020097828A1 (zh) * 2018-11-14 2020-05-22 深圳市欢太科技有限公司 回声消除方法、延时估计方法、装置、存储介质及设备
CN111402910B (zh) * 2018-12-17 2023-09-01 华为技术有限公司 一种消除回声的方法和设备
CN111383655B (zh) * 2018-12-29 2023-08-04 嘉楠明芯(北京)科技有限公司 一种波束形成方法、装置及计算机可读存储介质
CN110648679B (zh) * 2019-09-25 2023-07-14 腾讯科技（深圳）有限公司 回声抑制参数的确定方法和装置、存储介质及电子装置
CN110634496B (zh) * 2019-10-22 2021-12-24 广州视源电子科技股份有限公司 一种双讲检测方法、装置、计算机设备和存储介质
KR20210062475A (ko) * 2019-11-21 2021-05-31 삼성전자주식회사 전자 장치 및 그 제어 방법
US11122160B1 (en) * 2020-07-08 2021-09-14 Lenovo (Singapore) Pte. Ltd. Detecting and correcting audio echo
TWI757954B (zh) * 2020-11-05 2022-03-11 宏碁股份有限公司 會議終端及用於會議的多裝置協調方法
CN113223546A (zh) * 2020-12-28 2021-08-06 南京愔宜智能科技有限公司 一种音视频会议系统及用于该音视频会议系统的回音抵消装置

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101179294A (zh) * 2006-11-09 2008-05-14 爱普拉斯通信技术（北京）有限公司 自适应回声消除器及其回声消除方法
CN101346895A (zh) * 2005-10-26 2009-01-14 日本电气株式会社 回声抑制方法及设备
CN101826892A (zh) * 2009-03-03 2010-09-08 冲电气工业株式会社 回声消除器
US20110150067A1 (en) * 2009-12-17 2011-06-23 Oki Electric Industry Co., Ltd. Echo canceller for eliminating echo without being affected by noise
US20110317564A1 (en) * 2010-06-29 2011-12-29 Fadi Saibi High-Speed Ethernet Transceiver Calibration with Echo Canceller Reuse

Family Cites Families (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE602005013362D1 (de) * 2004-01-29 2009-04-30 Nxp Bv Echolöscher mit durch störungspegel gesteuerter schrittgrösse
US7643630B2 (en) * 2004-06-25 2010-01-05 Texas Instruments Incorporated Echo suppression with increment/decrement, quick, and time-delay counter updating
DE102004033866B4 (de) * 2004-07-13 2006-11-16 Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. Konferenz-Endgerät mit Echoreduktion für ein Sprachkonferenzsystem
US7539300B1 (en) * 2005-06-11 2009-05-26 Mindspeed Technologies, Inc. Echo canceller with enhanced infinite and finite ERL detection
CN101321201B (zh) 2007-06-06 2011-03-16 联芯科技有限公司 回声消除装置、通信终端及确定回声时延的方法
US7809129B2 (en) 2007-08-31 2010-10-05 Motorola, Inc. Acoustic echo cancellation based on noise environment
US8144862B2 (en) 2008-09-04 2012-03-27 Alcatel Lucent Method and apparatus for the detection and suppression of echo in packet based communication networks using frame energy estimation
CN201294570Y (zh) 2008-10-31 2009-08-19 比亚迪股份有限公司 一种用于电话会议系统的回声消除装置
CN101859652B (zh) 2009-04-08 2013-03-13 深圳富泰宏精密工业有限公司 按键装置
JP5347794B2 (ja) 2009-07-21 2013-11-20 ヤマハ株式会社 エコー抑圧方法およびその装置
US8625776B2 (en) 2009-09-23 2014-01-07 Polycom, Inc. Detection and suppression of returned audio at near-end
CN102316231B (zh) 2010-07-08 2014-03-12 杭州华三通信技术有限公司 一种回声消除的方法和装置
CN102347785B (zh) 2010-07-23 2013-09-11 联芯科技有限公司 一种回声消除方法及装置
CN102185991A (zh) 2011-03-01 2011-09-14 杭州华三通信技术有限公司 回声消除方法、系统和装置
EP2575375B1 (en) * 2011-09-28 2015-03-18 Nxp B.V. Control of a loudspeaker output
CN103888630A (zh) * 2012-12-20 2014-06-25 杜比实验室特许公司 用于控制声学回声消除的方法和音频处理装置

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101346895A (zh) * 2005-10-26 2009-01-14 日本电气株式会社 回声抑制方法及设备
CN101179294A (zh) * 2006-11-09 2008-05-14 爱普拉斯通信技术（北京）有限公司 自适应回声消除器及其回声消除方法
CN101826892A (zh) * 2009-03-03 2010-09-08 冲电气工业株式会社 回声消除器
US20110150067A1 (en) * 2009-12-17 2011-06-23 Oki Electric Industry Co., Ltd. Echo canceller for eliminating echo without being affected by noise
US20110317564A1 (en) * 2010-06-29 2011-12-29 Fadi Saibi High-Speed Ethernet Transceiver Calibration with Echo Canceller Reuse

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP2849351A4

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104778950A (zh) * 2014-01-15 2015-07-15 华平信息技术股份有限公司 一种基于回声消除的麦克风信号延时补偿控制方法
CN110956975A (zh) * 2019-12-06 2020-04-03 展讯通信（上海）有限公司 回声消除方法及装置

Also Published As

Publication number Publication date
EP2849351A4 (en) 2015-12-23
CN103391381A (zh) 2013-11-13
US20150124986A1 (en) 2015-05-07
EP2849351A1 (en) 2015-03-18
US9443528B2 (en) 2016-09-13
CN103391381B (zh) 2015-05-20
EP2849351B1 (en) 2017-05-03

Similar Documents

Publication Publication Date Title
WO2013166761A1 (zh) 回声消除方法及装置
CA2414972C (en) Gain control method for acoustic echo cancellation and suppression
US9082389B2 (en) Pre-shaping series filter for active noise cancellation adaptive filter
US8229147B2 (en) Hearing assistance devices with echo cancellation
CN103077726B (zh) 用于线性声学回声消除系统的预处理和后处理
US5390244A (en) Method and apparatus for periodic signal detection
WO2005125168A1 (en) Echo canceling apparatus, telephone set using the same, and echo canceling method
EP0853844B1 (en) Echo cancelling system for digital telephony applications
CN110956975B (zh) 回声消除方法及装置
JP5086769B2 (ja) 拡声通話装置
CN106571147B (zh) 用于网络话机声学回声抑制的方法
WO2021114779A1 (zh) 基于双端发声检测的回声消除方法、装置及系统
EP1786191A1 (en) Acoustic echo canceller
US20060198511A1 (en) Fast echo canceller reconvergence after TDM slips and echo level changes
US7539300B1 (en) Echo canceller with enhanced infinite and finite ERL detection
US8369511B2 (en) Robust method of echo suppressor
US20120140918A1 (en) System and method for echo reduction in audio and video telecommunications over a network
KR102172608B1 (ko) 동시통화에 강인한 심층학습 기반 음향반향 제거 장치 및 그 방법
JP5963077B2 (ja) 通話装置
Fukui et al. Acoustic echo canceller software for VoIP hands-free application on smartphone and tablet devices
JP2004274683A (ja) エコーキャンセル装置、エコーキャンセル方法、プログラムおよび記録媒体
EP1341365A1 (en) Method and arrangement for processing a speech signal
Leblanc et al. Performance improvement of acoustic ECHO cancellers in the presence of external volume change
JP2009302984A (ja) 音声通信装置および音声通信方法

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: 12876306

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

WWE Wipo information: entry into national phase

Ref document number: 14400289

Country of ref document: US

REEP Request for entry into the european phase

Ref document number: 2012876306

Country of ref document: EP

WWE Wipo information: entry into national phase

Ref document number: 2012876306

Country of ref document: EP