JP4829816B2 - Echo prevention device and program - Google Patents

Description

  The present invention relates to an echo prevention device and a program.

  In recent years, for example, in a communication device such as a mobile phone or a hands-free phone to which an earphone microphone is connected, an echo prevention device for preventing an echo caused by acoustic coupling or electrical reflection on a circuit from a speaker is incorporated. There is something that is.

  For example, Patent Document 1 discloses an echo prevention device that cancels echoes with high accuracy using digital processing. FIG. 4 is a diagram illustrating an example of an echo prevention apparatus using the DSP 100. As shown in the figure, an analog signal indicating voice transmitted from the other party using a mobile phone or the like is input to the AD converter 101. The signals digitally converted by the AD converter 101 are subjected to convolution processing by the FIR filters 102 and 103 in the DSP 100 based on the respective filter coefficients, and are output. A signal output from the FIR filter 102 is input to the DA converter 104. The signal analog-converted by the DA converter 104 is output to the earphone microphone via the input / output terminal 105 and is input to one terminal of the differential amplifier circuit 106. A signal output from the FIR filter 103 is input to the DA converter 107. The signal output from the DA converter 107 is input to the other terminal of the differential amplifier circuit 106. The signal output from the differential amplifier circuit 106 is converted into a digital signal by the AD converter 108 and input to the DSP 100.

  Here, the DSP 100 acquires an impulse response from the DA converter 104 to the AD converter 108 based on the output of the AD converter 108 when the impulse is output to the DA converter 104. Further, the DSP 100 acquires an impulse response from the DA converter 107 to the AD converter 108 based on the output of the AD converter 108 when the impulse is output to the DA converter 107. Then, by appropriately setting the filter coefficients of the FIR filters 102 and 103 based on these impulse responses, the echo due to the signal input to the AD converter 101 can be removed by the differential amplifier circuit 106.

The signal output from the AD converter 108 is output by subtracting the output signal of the adaptive filter 111 in the subtracting unit 110 in the DSP 100. The adaptive filter 111 adaptively changes the filter coefficient so that the output signal from the subtracting unit 110 is equal to or lower than a predetermined level in a state in which an audio signal is being transmitted from the other side, whereby the differential amplifier circuit 106. Echoes that could not be removed with are removed. The digital signal output from the subtracting unit 110 is output from the DSP 100, converted to an analog signal by the DA converter 109, and output as an output signal of the echo prevention device.
JP 2006-304260 A

  As described above, in the echo prevention device, after the echo is removed by the differential amplifier circuit 106, the echo is further removed in the DSP 100, but the echo is not completely removed. An echo may remain. However, for the other party's speaker, the echo that returns almost simultaneously with the utterance often has no sense of incongruity.

  By the way, in the echo prevention apparatus, a large noise may be superimposed on a signal transmitted to the other party due to the influence of noise from the earphone microphone 20 or electrical noise of the circuit. As a countermeasure for removing such noise, it is conceivable to output a signal output from the subtracting unit 110 through a noise canceller.

  However, since noise cancellers generally store input data to some extent, and then perform noise cancellation processing on the stored data and output it, a delay corresponding to the storage amount and processing amount occurs. It becomes. Therefore, even if the echo output from the signal output from the subtracting unit 110 is slight, the other party's speaker can feel a sense of discomfort because the echo is heard a little after the utterance.

  Therefore, an object of the present invention is to provide an echo prevention apparatus and program capable of reducing a sense of incongruity due to echo.

  An echo prevention apparatus according to the present invention for solving the above-described problems is provided with a first input signal and a second input signal including an echo generated by the first input signal, and the first input signal is used based on the first input signal. An echo removal unit that outputs a signal obtained by removing or attenuating the echo included in two input signals; a noise removal unit that outputs a signal obtained by removing or attenuating noise contained in the signal output from the echo removal unit; Based on the signal level of the echo included in the signal output from the echo removal unit, when the signal level is equal to or higher than a predetermined level, the signal output from the echo removal unit is output as an output signal, and the signal An output control unit that outputs the signal output from the noise removal unit as an output signal when the level is less than the predetermined level.

  In the program of the present invention, a first input signal and a second input signal including an echo due to the first input signal are input to the CPU, and the second input signal is input to the CPU based on the first input signal. Included in the function to output a signal with the echo removed or attenuated, the function to output the signal with the noise removed or attenuated, and the signal with the echo removed or attenuated When the signal level is equal to or higher than a predetermined level, the signal from which the echo is removed is output as an output signal, and when the signal level is lower than the predetermined level, the noise is removed or attenuated. And a function of outputting a signal as an output signal.

  It is possible to provide an echo prevention device and a program capable of reducing a sense of incongruity due to echo.

== Overall structure ==
First, the configuration of an echo prevention apparatus according to an embodiment of the present invention will be described. FIG. 1 is a block diagram showing the configuration of the echo prevention apparatus of this embodiment. The echo prevention apparatus includes a digital signal processor (DSP) 3, AD converters 4 and 5, DA converters 6 to 8, amplifier circuits 9 to 11, a differential amplifier circuit 12, and an input / output terminal 13. . Outside the echo prevention device, an earphone microphone 20 connected to the input / output terminal 13, a CPU (Central Processing Unit) 25 that controls the echo prevention device in an integrated manner, and a memory 26 that can be read and written by the CPU 25 are provided. ing.

  The earphone microphone 20 has a speaker function that generates sound by vibrating a diaphragm (not shown) based on an audio signal input from the input / output terminal 13. The earphone microphone 20 also has a microphone function for generating an audio signal by replacing the vibration of the eardrum when the person wearing the earphone microphone 20 emits sound with the vibration of the diaphragm (for example, JP 2003-9272 A). The audio signal generated by the earphone microphone 20 is input to the echo prevention device via the input / output terminal 13. The signal output to the earphone microphone 20 via the input / output terminal 13 is reflected and input from the input / output terminal 13 to the echo prevention device. Here, the reflected signal is, for example, a signal returned through the earphone microphone 20 or a sound output from the earphone microphone 20 reflected in the ear, and the reflected sound is converted into an audio signal by the earphone microphone 20. The converted signal or the like. The input / output terminal 13 does not exclusively input / output output signals and input signals. For example, the input / output terminal 13 may input / output an output signal and an input signal at the same time.

  The DSP 3 includes input terminals 30 and 31, output terminals 32 to 34, a DSP core 40, a RAM (Random Access Memory) 41, and a ROM (Read Only Memory) 42. The DSP 3 includes FIR filters 50 and 51, an impulse response acquisition unit 52, a filter coefficient setting unit 53, an echo canceller (EC) 54, a noise canceller (NC) 55, and a noise canceller control unit (NC control unit) 56. . The FIR filters 50 and 51, the impulse response acquisition unit 52, the filter coefficient setting unit 53, the echo canceller 54, the noise canceller 55, and the NC control unit 56 execute a program stored in the RAM 41 or the ROM 42 by the DSP core 40. It is realized by. The filter coefficients of the FIR filters 50 and 51 are stored in the RAM 41.

  The FIR filter 50 outputs a digital signal obtained by performing a convolution operation process on the input digital signal based on the filter coefficient of the FIR filter 50 to the output terminal 32 via the echo canceller 54. At the same time, the FIR filter 51 outputs, to the output terminal 33, a digital signal obtained by performing convolution operation processing on the input digital signal based on the filter coefficient of the FIR filter 51.

  The differential amplifier circuit 12 outputs a signal obtained by amplifying the difference between the analog signal input to the + input terminal and the analog signal input to the − input terminal, so that the output signal of the amplifier circuit 9 and the amplifier circuit A signal (echo) obtained by combining the output signal of 9 and the signal reflected and returned by the earphone microphone 20 is removed or attenuated by the differential amplifier circuit 12 using the output signal of the amplifier circuit 11. When the differential amplifier circuit 12 cannot completely remove the echo, a signal including the attenuated echo is output. Here, the signal including the attenuated echo refers to a signal in which the echo signal level is lowered although the echo is not completely removed.

  The echo canceller 54 (echo removal unit) performs echo removal processing on the signal input from the input terminal 31 and outputs the result. The echo canceller 54 is used to remove or attenuate an echo included in a signal output from the differential amplifier circuit 12, that is, an echo that cannot be removed by the differential amplifier circuit 12. In the echo canceller 54, if the echo cannot be completely removed, a signal including an attenuated echo is output.

  The noise canceller 55 (noise removal unit) performs noise removal processing on the signal output from the echo canceller 54 and outputs the result. The noise canceller 55 is, for example, “Y. Ephraim, D. Malah,“ Speech enhancement using a minimum mean-square error short-time spectral amplitude estimator ”, IEEE Trans. On ASSP, vol. ASSP-32, No. 6, pp. .1109-1121, 1984 ”may be employed. In the noise removal processing of the noise canceller 55, for example, a predetermined amount of input signal is accumulated, a signal-to-noise ratio is calculated for the accumulated data for each frequency band, a noise suppression coefficient is determined based on this signal-to-noise ratio, The noise is removed by multiplying the amplitude component of the input signal spectrum by this coefficient in the frequency domain. The noise removal processing in the noise canceller 55 can be turned on / off under the control of the NC control unit 56. When the noise removal processing is off, the signal output from the echo canceller 54 is output to the output terminal 34 without delay. Become.

  The NC control unit 56 (output control unit) controls on / off of noise removal processing in the noise canceller 55 according to the signal level of the echo remaining in the signal output from the echo canceller 54. For example, when the signal level of the echo remaining in the signal output from the echo canceller 54 is equal to or higher than a predetermined level, when the echo is delayed by the noise canceller 55 and transmitted to the other party, the other party's speaker feels uncomfortable. There is a risk that. Therefore, in such a case, the noise canceller control unit 56 outputs the signal output from the echo canceller 54 to the output terminal 34 without delaying by turning off the noise removal processing in the noise canceller 55. Conversely, if the signal level of the echo remaining in the signal output from the echo canceller 54 is less than a predetermined level, even if the echo is delayed and transmitted to the other party by the noise canceller 55, the other party's speaker feels uncomfortable. There is a high possibility of not having. Therefore, in such a case, the noise canceller control unit 56 turns on the noise removal processing in the noise canceller 55 to output a signal obtained by performing noise removal processing on the signal output from the echo canceller 54 to the output terminal 34. To do.

  The impulse response acquisition unit 52 includes an impulse response from the input terminal 31 when an impulse is generated in the output of the FIR filter 50, and an impulse response from the input terminal 31 when an impulse is generated in the output of the FIR filter 51. To get. The filter coefficient setting unit 53 outputs a signal (echo) that combines the output signal of the amplifier circuit 9 and the signal that is returned from the output signal of the amplifier circuit 9 after being reflected through the earphone microphone 20. The filter coefficients of the FIR filters 50 and 51 are set based on the impulse response acquired by the impulse response acquisition unit 52 so that the signal is removed or attenuated by the operation amplifier circuit 12.

  In such an echo prevention device, the audio signal input to the AD converter 4 is output to the earphone microphone 20 via the input / output terminal 13, and the diaphragm of the earphone microphone 20 vibrates to output sound. The echo generated by the audio signal input to the AD converter 4 is removed or attenuated by the differential amplifier circuit 12 and the echo canceller 54. When a person wearing the earphone microphone 20 utters a sound, the diaphragm of the earphone microphone 20 vibrates to generate a sound signal, and the generated sound signal is input via the input / output terminal 13, Sent to the side. The acquisition of the impulse response and the setting of the filter coefficient can be performed by a method similar to the method disclosed in Japanese Patent Application Laid-Open No. 2006-304260, for example.

== On / off control of noise canceller ==
Next, details of the on / off control of the noise canceller 55 in the present embodiment will be described. FIG. 2 is a block diagram showing an example of the peripheral configuration of the noise canceller 55 of the present embodiment. The echo canceller 54 includes a subtracting unit 72, an adaptive filter (ADF: Adaptive Digital Filter) 73, and an adaptive filter control unit (ADF control unit) 74. The DSP 3 is further provided with call detection units 70 and 71.

  A signal output from the FIR filter 50 is input to the call detection unit 70. The call detection unit 70 detects whether or not a voice signal is transmitted from the other party based on the signal output from the FIR filter 50. For example, if the signal level of the signal output from the FIR filter 50 is equal to or higher than a predetermined level, the call detecting unit 70 determines that a voice signal is being transmitted from the other side, and if the signal level is lower than the predetermined level, the other side. It is determined that no audio signal has been transmitted. The signal output from the FIR filter 50 is subjected to detection processing by the call detection unit 70 and is output to the output terminal 32.

  The call detection unit 71 receives a signal output from the subtraction unit 72. The call detector 71 detects the speech state of the wearer of the earphone microphone 20 based on the signal output from the subtractor 72. For example, the call detection unit 71 determines that the wearer of the earphone microphone 20 is speaking if the signal level of the signal output from the subtraction unit 72 is equal to or higher than a predetermined level, and if less than the predetermined level, the earphone microphone 20 It is determined that the wearer is not speaking. The signal output from the subtracting unit 72 is detected by the call detecting unit 71 and output to the noise canceller 55.

  The subtraction unit 72 generates a signal y (n) (pseudo echo signal) output from the adaptive filter 73 from a signal z (n) (echo signal: second input signal) including an echo input from the input terminal 31. The subtracted signal e (n) (removal error signal) is output.

  The adaptive filter 73 receives the signal x (n) (reference signal: first input signal) output from the FIR filter 50 and the output signal e (n) of the subtractor 72. The adaptive filter 73 receives the audio signal from the other side, and the signal e (n) output from the subtracting unit 72 is not more than a predetermined level when the wearer of the earphone microphone 20 is not speaking. The filter coefficients are adaptively changed so that The configuration of the adaptive filter 73 and the filter coefficient setting operation can be equivalent to the configuration and operation of the adaptive filter disclosed in, for example, Japanese Patent Application Laid-Open No. 2006-304260.

  Based on the detection results of the call detection units 70 and 71, the ADF control unit 74 filters the adaptive filter 73 in a state where a voice signal has been transmitted from the other party and the wearer of the earphone microphone 20 is not speaking. The adaptive update process of the coefficient is executed.

  Based on the detection results of the call detection units 70 and 71, the NC control unit 56 outputs an audio signal from the other party and outputs from the subtraction unit 72 when the wearer of the earphone microphone 20 is not speaking. The signal level of the signal e (n) to be detected is detected. The signal level of the signal e (n) detected in this state indicates the level of echo (attenuated echo) that remains without being removed by the echo canceller 54. When the signal level of the signal e (n) is equal to or higher than the predetermined level, the NC control unit 56 turns off the noise cancellation processing in the noise canceller 55, and when the signal level of the signal e (n) is lower than the predetermined level, The noise canceling process in the noise canceller 55 is turned on. Here, when the signal level of the signal e (n) is less than the predetermined level, the echo is completely removed, and the case where the signal level of the signal e (n) is zero is also included. The signal level of the detected signal e (n) indicates the level of the signal e (n), such as the power in a predetermined period, the maximum amplitude in the predetermined period, and the sum of absolute values of the amplitudes in the predetermined period. Anything is possible.

  FIG. 3 is a flowchart showing an example of processing for controlling on / off of the noise canceller 55 in the present embodiment. First, when the power is turned on and the echo prevention apparatus is activated (S301), the NC control unit 56 turns off the noise cancellation processing in the noise canceller 55. Thereafter, a call between the wearer of the earphone microphone 20 and the other party is started (S303). Immediately after activation of the echo prevention device, the adaptive update process of the filter coefficient in the adaptive filter 73 is not completed, and the signal e (n) output from the echo canceller 54 includes a certain amount of echoes. In many cases. For this reason, the noise canceller 55 is turned off immediately after activation of the echo prevention device, so that an echo caused by a delay due to the noise canceling process is not sent to the other party, and the uncomfortable feeling due to the delayed echo can be suppressed.

  After the start of the call, the NC control unit 56 echoes in a state in which a voice signal has been transmitted from the other party based on the detection results of the call detection units 70 and 71 and the wearer of the earphone microphone 20 is not speaking. The signal level E of the signal e (n) output from the canceller 54 is detected (S304). Then, the NC control unit 56 compares the detected signal level E with a threshold value Sh indicating a predetermined level (S305).

  When the signal level E of the signal e (n) is equal to or higher than the threshold value Sh (S305: E ≧ Sh), the NC control unit 56 turns off the noise cancellation processing in the noise canceller 55 (S306). That is, when the signal e (n) output from the echo canceller 54 includes a certain amount of echo, the noise canceller 55 is turned off, so that the echo delayed by the noise canceling process is not detected. It is not sent to the side, and the uncomfortable feeling due to the delayed echo can be suppressed.

  When the signal level E of the signal e (n) is less than the threshold Sh (S305: E <Sh), the NC control unit 56 turns on the noise cancellation processing in the noise canceller 55 (S307). That is, when the echo included in the signal e (n) output from the echo canceller 54 is very small, the noise canceller 55 is turned on, and a clear signal from which noise has been removed is transmitted to the other party. The In this state, since the echo included in the signal e (n) is very small, even if an echo delayed by the noise cancellation processing in the noise canceller 55 is transmitted to the other side, the possibility that the other side has a sense of incongruity is reduced.

  The NC control unit 56 confirms whether the call is continued between the wearer of the earphone microphone 20 and the other party based on the detection results of the call detection units 70 and 71 (S308). When the call is in progress (S308: YES), the NC control unit 56 repeatedly executes the on / off processing (S304 to S307) of the noise canceller 55 based on the signal level of the signal e (n). When the call ends (S308: NO), the process ends. When the filter coefficient of the adaptive filter 73 is updated to an optimum value and the signal level of the signal e (n) once becomes less than the threshold Sh, the noise canceller 55 is fixed to be on, and thereafter the noise canceller 55 It is also possible not to switch on / off.

  The embodiment of the present invention has been described above. As described above, in the echo prevention apparatus of this embodiment, when the signal level of the echo included in the signal e (n) output from the echo canceller 54 is equal to or higher than a predetermined level, the signal output from the echo canceller 55 is The noise canceller 55 outputs the signal without performing the noise canceling process. On the other hand, when the signal level of the echo included in the signal e (n) output from the echo canceller 54 is less than a predetermined level, the signal output from the echo canceller 55 is subjected to noise cancellation processing in the noise canceller 55. It is output after. That is, if the signal level of the echo included in the signal e (n) is high enough that the other party's speaker will feel uncomfortable if the echo is transmitted with a delay, no delay due to the noise cancellation process is caused. Thus, the uncomfortable feeling due to the echo can be reduced. On the other hand, if the signal level of the echo contained in the signal e (n) is small enough that the other party's speaker does not feel uncomfortable even if the echo is transmitted with a delay, a noise cancellation process is performed to The signal transmitted to the side can be made clearer.

  Further, in the echo preventing apparatus of the present embodiment, the echo canceller 54 is configured using the adaptive filter 73 that gradually reduces the echo by adaptively changing the filter coefficient. For this reason, it takes some time until the filter coefficient of the adaptive filter 73 is updated to an appropriate value after the echo prevention apparatus is activated. Therefore, it is particularly effective to turn on / off the noise cancellation processing according to the signal level of the echo included in the signal e (n) output from the echo canceller 54.

  Further, in the echo preventing apparatus of the present embodiment, the noise canceling process in the noise canceller 54 is turned off at the start-up when the power is turned on. In other words, since the filter coefficient of the adaptive filter 73 is not updated to an appropriate value when the echo prevention device is activated, by turning off the noise cancellation processing, an echo that is uncomfortable for the other speaker Can be prevented from being transmitted.

  The embodiments and application examples described above are intended to facilitate understanding of the present invention, and are not intended to limit the present invention. The present invention can be changed and improved without departing from the gist thereof, and the present invention includes equivalents thereof.

  For example, in this embodiment, the impulse response obtained by generating the impulse is set as the filter coefficient of the FIR filter, but the signal used for setting the filter coefficient is not limited to the impulse. For example, the filter coefficient of the FIR filter may be set based on the response signal obtained when the step signal is generated.

  Moreover, although this embodiment demonstrated the echo prevention apparatus used with the earphone microphone 20, an echo prevention apparatus is not only the earphone microphone 20, but an apparatus in which an echo is sent back to the other party side by an audio signal transmitted from the other party side. Applicable if available. For example, when the telephone is used in the hands-free mode, the echo prevention device can be used even in a configuration in which the sound output from the telephone speaker is input from the telephone microphone and the echo is transmitted to the other party. It is possible to apply.

It is a block diagram which shows the structure of the echo prevention apparatus of this embodiment. It is a block diagram which shows an example of the periphery structure of the noise canceller of this embodiment. It is a flowchart which shows an example of the process which controls ON / OFF of the noise canceller in this embodiment. It is a figure which shows an example of the echo prevention apparatus using DSP.

Explanation of symbols

3 DSP
4,5 AD converter 6-8 DA converter 9-11 Amplifier circuit 12 Differential amplifier circuit 13 Input / output terminal 20 Earphone microphone 25 CPU
26 Memory 30, 31 Input terminal 32-34 Output terminal 40 DSP core 41 RAM
42 ROM
50, 51 FIR filter 52 Impulse response acquisition unit 53 Filter coefficient setting unit 54 Echo canceller 55 Noise canceller 56 Noise canceller control unit 70, 71 Call detection unit 72 Subtraction unit 73 Adaptive filter 74 Adaptive filter control unit

Claims (4)

A first input signal and a second input signal including an echo generated by the first input signal are input, and a signal obtained by removing or attenuating the echo included in the second input signal based on the first input signal. An echo canceler to output,
A noise removing unit that outputs a signal obtained by removing or attenuating noise included in the signal output from the echo removing unit;
Based on the signal level of the echo included in the signal output from the echo removal unit, when the signal level is equal to or higher than a predetermined level, the signal output from the echo removal unit is output as an output signal, and the signal An output control unit that outputs the signal output from the noise removing unit as an output signal when the level is less than the predetermined level;
An echo prevention apparatus comprising: The echo prevention device according to claim 1,
The echo removing unit
An adaptive filter to which the first input signal is input;
A subtractor that subtracts and outputs the output signal of the adaptive filter from the second input signal;
With
Based on the first input signal and the output signal of the subtraction unit, the adaptive filter uses a filter coefficient to make the output signal of the subtraction unit a signal obtained by removing or attenuating the echo included in the second input signal. Can be updated adaptively,
Echo prevention device characterized by. The echo prevention device according to claim 2,
The output control unit
When starting the echo prevention device, outputting the signal output from the echo removal unit as the output signal,
Echo prevention device characterized by. CPU
A first input signal and a second input signal including an echo generated by the first input signal are input, and a signal obtained by removing or attenuating the echo included in the second input signal based on the first input signal. A function to output,
A function of outputting a signal obtained by removing or attenuating noise included in a signal obtained by removing or attenuating the echo;
Based on the signal level of the echo contained in the signal from which the echo is removed or attenuated, if the signal level is equal to or higher than a predetermined level, the signal from which the echo is removed or attenuated is output as an output signal, and the signal level is the predetermined level When the level is less than the level, a function of outputting the signal from which the noise is removed or attenuated as an output signal
A program to realize

Images