JP5051235B2 - Echo suppression system, echo suppression method, echo suppression program, echo suppression device, and sound output device - Google Patents

Description

The present invention relates to a sound output device that outputs sound based on a sound signal, a sound input device to which sound is input, and an echo based on the sound output by the sound output device from the sound input by the sound input device. In an echo suppression system comprising an echo suppression device for suppression, an echo suppression method applied to the echo suppression system, an echo suppression program for realizing an echo suppression device used in the echo suppression system, and the echo suppression system about the echo suppressor and a sound output equipment used.

  For example, in speech recognition in a conventional car navigation system, the car navigation system is muted by pressing the utterance switch before the speaker speaks, and the sound output from the speaker of the car navigation system is echoed to the microphone. This prevents wraparound and misrecognition of voice. On the other hand, there is a request from the user to perform an operation by voice recognition without muting the output sound. In order to meet such a demand, an echo suppression method for suppressing the sound that circulates from the speaker to the microphone is required. In addition, in a car audio system, that is, in a car audio system, a plurality of speakers are used to support multi-channel, so that it corresponds to an echo that circulates from a multi-channel car audio system to a microphone used for voice recognition or the like. An echo suppression method is required.

  Therefore, an echo suppression apparatus has been proposed in which an echo suppression method corresponding to a monaural channel audio system is applied to a multi-channel audio system. FIG. 1 is a schematic diagram showing a configuration of a conventional first echo suppressor. In FIG. 1, reference numeral 1000 denotes an echo suppression apparatus. The echo suppression apparatus 1000 outputs a plurality of channels of sound signals output from the multichannel audio 2000 to a plurality of speakers 2001, 2001,. ,... Output based on the respective sound signals. Then, the echo suppression apparatus 1000 removes echoes based on the output sound of a plurality of channels from the sound received by the microphone 2002. The echo is removed from the observed sound signal y (t) based on the received sound by using the reference sound signals x1 (t),..., Xn (t) based on the sounds of the plurality of channels (n) to be output on the respective channels. , Xn (t) corresponding to the reference sound signals x1 (t),..., Xn (t) are suppressed by suppressing the corresponding echoes by a plurality of suppression mechanisms (echo cancellers) 1001-1,.

  An echo suppressor that applies an echo suppression method corresponding to a monaural channel audio system to a multi-channel audio system has also been proposed for forms other than FIG. FIG. 2 is a schematic diagram showing a configuration of a conventional second echo suppression device. The second echo suppressor 1000 generates an added reference sound signal x (t) by an adding mechanism 1002 that adds reference sound signals x1 (t),..., Xn (t) based on sounds of a plurality of channels. The suppression mechanism 1001 suppresses echoes from the observed sound signal y (t) based on the added reference sound signal x (t).

  FIG. 3 is a functional block diagram showing a functional configuration of a suppression mechanism 1001 provided in a conventional echo suppression device. The suppression mechanism 1001 includes a detection unit 10010 that detects a state of double talk that the speaker is uttering and a state of single talk that is not uttered, and a filter coefficient required for echo estimation by adaptive processing using a learning identification method. A filter coefficient updating unit 10011 for updating the signal, a linear FIR filter unit 10012 for estimating the echo signal x ′ (t) from the reference sound signal x (t) by a several hundredth-order product-sum operation using the filter coefficient, A subtractor 10013 that outputs a suppression result r (t) obtained by performing echo suppression by subtracting the echo signal x ′ (t) from the observation sound signal y (t). The detection unit 10010 detects a single talk state and a double talk state based on the intensity change of the suppression result r (t), and stops the update of the filter coefficient of the filter coefficient update unit 10011 in the double talk state. . The filter coefficient update unit 10011 calculates a filter coefficient based on the suppression result r (t).

The echo suppression apparatus 1000 shown in FIG. 1 includes the suppression mechanism 1001 shown in FIG. 3 for each reference sound signal x1 (t),..., Xn (t). Such an echo suppression method is disclosed in Patent Document 1, for example.
JP 2002-237769 A

  However, the method requiring a suppression mechanism corresponding to each channel as shown in FIG. 1 has problems such as an increase in cost and an increase in the size of the apparatus. In particular, when applied to a car navigation system having a large installation space restriction, the problem of enlargement is serious.

  Further, as shown in FIG. 2, when the reference sound signals are added to obtain a mono channel addition reference sound signal, there arises a problem that the residual echo that cannot be suppressed increases. This is because in the output of sound such as music by the multi-channel audio 2000, the reproduction sound and intensity of each speaker change independently, and therefore, if the echoes of a plurality of paths are learned by one adaptive process, the estimation accuracy is reduced. It is.

The present invention has been made in view of such circumstances, and a reference sound signal in which components of different frequency bands are passed through a sound signal of a plurality of channels and a plurality of sound signals of components of different frequency bands are added. In addition, by suppressing the echo included in the observation sound signal based on the observation sound signal obtained by converting the input sound, it is possible to suppress the echo corresponding to a plurality of channels with a single suppression mechanism. It is possible to suppress an increase in the size and size of the apparatus, and since an arbitrary frequency of the observation sound signal includes only one path echo based on the sound signal of one channel, it is possible to estimate the echo. Echo suppression system capable of increasing accuracy and suppressing residual echo, echo suppression method applied to the echo suppression system, used in the echo suppression system Echo suppressing program for realizing the echo suppressor which is directed to the provision of the echo suppressor and a sound output equipment used in the echo suppressing system.

  A first echo suppression system includes: a sound output device that outputs sound based on a sound signal; a sound input device that converts an input sound into a sound signal; and a sound input from the sound input device, In an echo suppression system comprising an echo suppression device that suppresses echo based on sound output from an output device, the sound output device includes a passing unit that allows a plurality of sound signals to pass through components in different frequency bands, and A plurality of sound output units that output sounds based on the plurality of sound signals that have passed through the passage unit, respectively, and an addition unit that generates a reference sound signal by adding the plurality of sound signals that have passed through the passage unit. The echo suppression device suppresses an echo included in the observation sound signal based on the input unit that receives the sound signal from the sound input device as an observation sound signal, and the observation sound signal and the reference sound signal. The above view It is a requirement to have a correcting unit for correcting a sound signal.

  The second echo suppression system is the first echo suppression system, wherein the passing unit includes a first conversion unit that converts a plurality of sound signals into components on the frequency axis, and a plurality of components that are converted into components on the frequency axis. A plurality of sound signals converted to components on the frequency axis through which the components of the different frequency bands are passed and the components on the frequency axis through which the components of the respective frequency bands are passed. It is a requirement to have a second conversion unit for converting into the above sound signal.

  The third echo suppression system is the second echo suppression system, wherein the plurality of sound signals through which the passing section passes components of different frequency bands are a first sound signal, a second sound signal, a first sound signal, and And / or a processed sound signal obtained by processing the second sound signal by a predetermined processing method, wherein the band-pass filter unit has a filter coefficient set for each sound signal, and the first filter coefficient for the first sound signal The second filter coefficient is set for the second sound signal, and the first filter coefficient and / or the processing filter coefficient based on the second filter coefficient corresponding to the processing method is set for the processed sound signal. And

  The fourth echo suppression system is the third echo suppression system, wherein the processed sound signal delays the chord signal, which is the sum of the first sound signal and the second sound signal, and the first sound signal by a predetermined time. The delayed first sound signal and the delayed second sound signal obtained by delaying the second sound signal for a predetermined time, and the processing filter coefficient is the sum of the first sound signal and the second sound signal with respect to the chord signal. It is a requirement that a filter coefficient based on the first filter coefficient is set for the delayed first sound signal, and a filter coefficient based on the second filter coefficient is set for the delayed second sound signal.

  According to a fifth echo suppression system, in any one of the first to fourth echo suppression systems, the passage section is configured to pass components in a frequency band divided at equal intervals.

  In a sixth echo suppression system, in any one of the first to fourth echo suppression systems, the passing unit allows a frequency band component divided so that logarithmic values are equally spaced to pass. Is a requirement.

  According to a seventh echo suppression system, in any one of the first to sixth echo suppression systems, the sound output device applies a frequency band component corresponding to a preset removal band to any sound signal. The correction unit performs correction for deriving a correction amount required for correcting the observed sound signal by filtering the reference sound signal with a filter coefficient set for each frequency. A filter unit; a coefficient update unit that calculates and updates a filter coefficient of the correction filter based on the corrected observation sound signal; and the coefficient update unit based on a component of the removal band of the observation sound signal It is a requirement to have an update availability determination unit that determines whether update is possible.

  An eighth echo suppression system according to any one of the first to seventh echo suppression systems, wherein the sound output device further compares amplitudes of a plurality of sound signals for each frequency band by a predetermined method. A comparator that determines whether or not a value indicating a comparison result between the amplitude of one sound signal having the maximum amplitude and the amplitudes of all other sound signals is equal to or greater than a predetermined value; The passage section is required to pass only the one sound signal for the frequency band determined by the comparison section to be equal to or greater than a predetermined value.

  A ninth echo suppression system according to any one of the first to seventh echo suppression systems, wherein the sound output device further compares amplitudes of a plurality of sound signals for each frequency band by a predetermined method. A comparator that determines whether or not a value indicating a comparison result between the amplitude of one sound signal having the maximum amplitude and the amplitudes of all other sound signals is equal to or greater than a predetermined value; The passage section is required to pass all sound signals for the frequency band determined by the comparison section to be equal to or greater than a predetermined value.

  According to a tenth echo suppression system, in any one of the first to ninth echo suppression systems, the passing section changes a frequency band component for each sound signal to be passed with time. As a requirement.

  An eleventh echo suppression method includes: a sound output device that outputs sound based on a sound signal; a sound input device that receives sound; and a sound input device that outputs the sound from the sound output device. In an echo suppression method using an echo suppressor that suppresses echo based on sound, the sound output device causes a plurality of sound signals to pass through components in different frequency bands, and in each of the different frequency bands. The sound output procedure for outputting sounds based on a plurality of sound signals through which the component has passed, the sound input procedure for converting the input sound into a sound signal by the sound input device, and the frequency different by the echo suppression device, respectively. An addition procedure for generating a reference sound signal by adding a plurality of sound signals through which band components have passed, an input procedure for inputting a sound signal from the sound input device as an observation sound signal, and the observation sound signal Based on the fine the reference sound signal, it is required for the containing and correction procedure for correcting the observation sound signal so as to suppress the echo contained in the observed sound signal.

  The twelfth echo suppression program cooperates with a sound output device that outputs a sound based on a sound signal and a sound input device that generates a sound signal based on an input sound, and the sound input by the sound input device In an echo suppression program for causing an echo suppression apparatus to execute a procedure for suppressing an echo based on the sound output from the sound output apparatus, the echo suppression apparatus includes components of different frequency bands output from the sound output apparatus. Based on a reference sound signal generated by adding a plurality of sound signals passed through and an observation sound signal which is a sound signal generated by the sound input device, echoes contained in the observation sound signal are suppressed. Therefore, it is necessary to execute a correction procedure for correcting the observation sound signal.

  The thirteenth echo suppressor cooperates with a sound output device that outputs sound based on the sound signal and a sound input device that generates sound signal based on the input sound, and the sound input by the sound input device In the echo suppressor for suppressing echo based on the sound output from the sound output device, a reference sound signal is obtained by adding a plurality of sound signals that have passed through components of different frequency bands output from the sound output device. Based on the observation sound signal and the reference sound signal, and an echo contained in the observation sound signal is suppressed, based on the addition part that generates the sound signal from the sound input device as an observation sound signal Therefore, it is necessary to have a correction unit that corrects the observation sound signal.

  The fourteenth sound output device includes a plurality of sound output units that output sound based on a plurality of sound signals, and the sound input device that generates a sound signal based on the input sound and the sound input device input A sound output device that cooperates with an echo suppression device that suppresses echo based on the sound output from the sound output portion from the generated sound, and has a passing portion that allows components of different frequency bands to pass through for each of the plurality of sound signals It is a requirement that the plurality of sound output units are configured to output sounds based on the plurality of sound signals that have passed through the passage unit.

  A fifteenth audio system includes: a sound output device that outputs sound based on a sound signal; a sound input device that converts an input sound into a sound signal; and the sound output from the sound input to the sound input device An audio system comprising: an echo suppression device that suppresses echo based on sound output from a device; and a control device that has a control unit that controls at least one of the sound output device, the sound input device, and the echo suppression device. The output device has a passage unit that allows components of different frequency bands to pass through a plurality of sound signals, and a plurality of sound output units that respectively output sounds based on the plurality of sound signals that have passed through the passage unit. An adder that generates a reference sound signal by adding a plurality of sound signals that have passed through the passage, and the echo suppression device is an input unit that receives the sound signal from the sound input device as an observation sound signal A correction unit that corrects the observation sound signal to suppress an echo included in the observation sound signal based on the observation sound signal and the reference sound signal, and the control unit that corrects the observation sound signal corrected by the correction unit. And an output unit that outputs the corrected observation sound signal that has been output to the input unit, and an audio signal based on the corrected observation sound signal that has been input from the input unit. A recognition unit that performs a recognition process, and the control unit performs control based on a recognition result of the recognition unit.

  A sixteenth navigation system includes a sound output device that outputs a sound based on a sound signal, a sound input device that converts an input sound into a sound signal, and the sound output from the sound input to the sound input device. In a navigation system including an echo suppression device that suppresses echo based on sound output from a device and a navigation device having a control unit that performs navigation processing, the sound output device has different frequency bands for a plurality of sound signals, respectively. And a plurality of sound output units for outputting sounds based on a plurality of sound signals that have passed through the passage unit, and adding a plurality of sound signals that have passed through the passage unit. An adder for generating a reference sound signal, wherein the echo suppressor includes an input unit that receives a sound signal from the sound input device as an observation sound signal, the observation sound signal, and the observation sound signal. A correction unit that corrects the observation sound signal to suppress an echo included in the observation sound signal based on the illuminating signal, and an output unit that outputs the corrected observation sound signal to the navigation device; The navigation device includes an input unit that receives the output of the corrected observation sound signal that has been output, and a recognition unit that performs speech recognition processing based on the corrected observation sound signal that has been input from the input unit. The control unit is required to perform navigation processing based on the recognition result by the recognition unit.

  The seventeenth moving body includes a sound output device that outputs a sound based on a sound signal, a sound input device that converts an input sound into a sound signal, and the sound output from the sound input to the sound input device. In a moving body including an echo suppression device that suppresses echo based on sound output from a device and a navigation device having a control unit that performs navigation processing, the sound output device has different frequency bands for a plurality of sound signals, respectively. And a plurality of sound output units for outputting sounds based on a plurality of sound signals that have passed through the passage unit, and adding a plurality of sound signals that have passed through the passage unit. The echo suppression device includes an adder that generates a reference sound signal, and the echo suppression device is based on an input unit that receives a sound signal from the sound input device as an observation sound signal, and the observation sound signal and the reference sound signal. The observed sound A correction unit that corrects the observation sound signal so as to suppress the echo included in the signal, and an output unit that outputs the corrected observation sound signal to the navigation device, and the navigation device outputs the correction An input unit to which a later observation sound signal is input; and a recognition unit that performs speech recognition processing based on the corrected observation sound signal input from the input unit; and the control unit is configured by the recognition unit It is a requirement to perform navigation processing based on the recognition result.

  First, second, fifth and sixth echo suppression systems, eleventh echo suppression method, twelfth echo suppression program, thirteenth echo suppression device, fourteenth sound output device, fifteenth audio system, In the sixteenth navigation system and the seventeenth moving body, since the sound signal is different for each frequency, only one path echo based on one sound signal is included for an arbitrary frequency of the observed sound signal. Therefore, since the estimation accuracy of echo can be increased, residual echo can be suppressed. In addition, it is not necessary to provide a correction unit for suppressing echoes for each of a plurality of sound signals, and one correction unit can suppress echoes caused by a plurality of sound signals. It is possible to prevent an increase in cost and an increase in size of the apparatus.

  In the third and fourth echo suppression systems, for example, a chord signal based on the sum of the first sound signal and the second sound signal, a delayed first sound signal obtained by delaying the first sound signal for a predetermined time, and a second sound signal are delayed. When a pseudo 5-channel stereo is realized using the processed sound signal such as the delayed second sound signal, the processed sound signal is used for the first filter coefficient and the second sound signal used for the first sound signal. By diverting the second filter coefficient and allocating the frequency band, it is possible to process three or more types of sound signals with two types of filters, so that the resources of the sound output device can be used efficiently. It becomes possible.

  In the seventh echo suppression system, for example, a frequency band corresponding to a human voice is set as a removal band, and based on a component of the removal band of the observation sound signal, whether the speaker is in a double talk state. Detecting whether it is in a single talk state that is not speaking, for example, if it is determined that it is in a double talk state, the update of the filter coefficient is stopped, thereby improving the recognition accuracy of the section containing speech Thus, it is possible to improve the accuracy of the voice recognition process.

  In the eighth and ninth echo suppression systems, one sound signal passes when the amplitude of one sound signal is greater than the amplitude of all other sound signals by a predetermined condition or more. Therefore, it is possible to output a sound with no sense of incongruity in a plurality of channels.

  In the tenth echo suppression system, even when there is a bias in the output of each sound signal, the specific sound signal does not continue to be blocked in a specific frequency band, so that a sound with no sense of incongruity is output in a plurality of channels. It is possible.

  The present application relates to a plurality of sound output units such as a speaker and the like based on a plurality of sound signals through which components of different frequency bands pass through a plurality of sound signals such as multi-channel audio. Respectively. A sound input device having a sound input unit such as a microphone converts an input sound into a sound signal. The echo suppressor is included in the observed sound signal based on a reference sound signal obtained by adding a plurality of sound signals having components of different frequency bands, and an observed sound signal that is a sound signal input from the sound input device. Disclosed is a technique for correcting an observed sound signal so as to suppress echoes generated.

  In this application, since the sound signal assigned for each frequency is different in this application, for any frequency of the observed sound signal, only the echo of one path based on one sound signal is included. Since the influence of an arbitrary frequency of the output sound appears at the same frequency of the input sound, it is possible to improve the estimation accuracy of the echo, so that the residual echo can be suppressed, etc. Has an excellent effect. In addition, in the present invention, it is not necessary to provide a correction unit for suppressing the echo for each of the plurality of sound signals, and the correction unit can suppress the echo due to the plurality of sound signals with one correction unit. As a result, it is possible to prevent an increase in cost and an increase in the size of the apparatus due to the increase in the number of devices.

  In the present application, when a processed sound signal obtained by processing the first sound signal, the second sound signal, the first sound signal and / or the second sound signal by a predetermined processing method is used as the plurality of sound signals, for example, the first sound signal When the signal and the second sound signal are two-channel stereo signals, a chord signal based on the sum of the first sound signal and the second sound signal, a delayed first sound signal obtained by delaying the first sound signal for a predetermined time, When the pseudo 5-channel stereo is realized using the processed sound signal such as the delayed second sound signal obtained by delaying the second sound signal, the first filter coefficient used for the first sound signal and the processed sound signal A technique for allocating a frequency band by diverting a second filter coefficient used for a second sound signal is disclosed.

  With this configuration, in the present application, since it is possible to perform processing on three or more types of sound signals with two types of filters, it is possible to effectively use the resources of the sound output device, and thus have excellent effects. .

  Further, in the present invention, in the sound output device, for example, a frequency band corresponding to a human voice is set as a removal band in advance, and a component of the frequency band corresponding to the removal band is not allowed to pass through any sound signal. To. Then, the echo suppressor detects whether the speaker is uttering a double talk state or not singing a single talk state based on the component of the observed sound signal removal band, When it is determined that the state is in a state, the filter coefficient of the correction filter unit for deriving the correction amount necessary for correcting the observation sound signal is calculated and updated by filtering with the filter coefficient set for each frequency, and doubled. Disclosed is a technique for stopping the update of the filter coefficient when it is determined that the state is a talk state.

  With this configuration, in the present application, there are excellent effects such as improving the recognition accuracy of the section including the speech and improving the speech recognition accuracy.

  Further, the present application compares the amplitudes of a plurality of sound signals by a predetermined method, and compares the amplitude of one sound signal having the maximum amplitude with the amplitudes of all other sound signals. For a frequency band in which the value indicating the value is determined to be greater than or equal to a predetermined value, a technique is disclosed in which all sound signals or only one sound signal having the maximum magnitude is passed.

  With this configuration, in the present application, for example, in the case of solo performance in an orchestra, when the amplitude of one sound signal is greater than or equal to a predetermined value relative to other sound signals, the one sound signal is blocked at the passage portion. As a result, it is possible to output an uncomfortable sound on a plurality of channels, and the effects are excellent.

  Furthermore, the present application discloses a technique for changing a frequency band component for each sound signal to be passed with time.

  With this configuration, in the present application, even when there is a bias in the output of each sound signal, since the specific sound signal is not continuously blocked in a specific frequency band, it is possible to output a sound that does not feel uncomfortable in multiple channels. It is possible to achieve an excellent effect.

It is a schematic diagram which shows the structure of the conventional 1st echo suppression apparatus. It is a schematic diagram which shows the structure of the conventional 2nd echo suppression apparatus. It is a functional block diagram which shows the function structure of the suppression mechanism with which the conventional echo suppression apparatus is provided. It is a block diagram which shows typically the structural example of the echo suppression system which concerns on Embodiment 1 of this invention. It is a functional block diagram which shows the structural example of the passage mechanism of the sound output device which concerns on Embodiment 1 of this invention. It is a graph which shows the filter coefficient of the passage filter part of the passage mechanism with which the sound output device concerning Embodiment 1 of the present invention is provided. It is a functional block diagram which shows the structural example of the suppression mechanism with which the echo suppression apparatus which concerns on Embodiment 1 of this invention is provided. It is a block diagram which shows the structural example of the sound processing apparatus which concerns on Embodiment 1 of this invention. It is a flowchart which shows an example of the sound output process of the sound output device which concerns on Embodiment 1 of this invention. It is a flowchart which shows an example of the sound input process of the sound input device which concerns on Embodiment 1 of this invention. It is a flowchart which shows an example of the echo suppression process of the echo suppression apparatus which concerns on Embodiment 1 of this invention. It is a flowchart which shows an example of the filter coefficient update process of the echo suppression apparatus which concerns on Embodiment 1 of this invention. It is a functional block diagram which shows the structural example of the passage mechanism of the sound output device which concerns on Embodiment 2 of this invention. It is a graph which shows the filter coefficient of the passage filter of the passage mechanism with which the sound output device concerning Embodiment 2 of the present invention is provided. It is a functional block diagram which shows the structural example of the passage mechanism of the sound output device which concerns on Embodiment 3 of this invention. It is a graph which shows the filter coefficient of the removal part of the passage mechanism with which the sound output device which concerns on Embodiment 3 of this invention is provided. It is a functional block diagram which shows the structural example of the suppression mechanism with which the echo suppression apparatus which concerns on Embodiment 3 of this invention is provided. It is a graph which shows the filter coefficient of the removal band pass filter part of the suppression mechanism with which the echo suppression apparatus which concerns on Embodiment 3 of this invention is provided. It is a flowchart which shows an example of the sound output process of the sound output device which concerns on Embodiment 3 of this invention. It is a flowchart which shows an example of the filter coefficient update process of the echo suppression apparatus which concerns on Embodiment 3 of this invention. It is a block diagram which shows typically the structural example of the echo suppression system which concerns on Embodiment 4 of this invention. It is a functional block diagram which shows the structural example of the passage mechanism of the sound output device which concerns on Embodiment 5 of this invention. It is a flowchart which shows an example of the sound output process of the sound output device which concerns on Embodiment 5 of this invention. It is explanatory drawing which shows notionally the example of the process of the passage mechanism with which the sound output device which concerns on Embodiment 5 of this invention is provided. It is a functional block diagram which shows the structural example of the passage mechanism of the sound output device which concerns on Embodiment 6 of this invention. It is a flowchart which shows an example of the sound output process of the sound output device which concerns on Embodiment 6 of this invention. It is a functional block diagram which shows the structural example of the passage mechanism of the sound output device which concerns on Embodiment 7 of this invention. It is a flowchart which shows an example of the coefficient change process of the sound output device which concerns on Embodiment 7 of this invention. It is explanatory drawing which shows the example of a time-dependent change of the filter coefficient of each band pass filter part of the pass mechanism with which the sound output device which concerns on Embodiment 7 of this invention is provided.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Echo suppression apparatus 10 1st input mechanism 11 2nd input mechanism 12 Suppression mechanism 120 Adder 121 Correction | amendment part 1210 Linear FIR filter part 1211 Subtraction part 1212 Filter coefficient update part 1213 Detection part 1214 Removal band pass filter part 13 Output mechanism 2 Sound Output device 20 Sound output mechanism 21 Sound signal generation mechanism 22 Passing mechanism 220 FFT conversion unit 221 Band pass filter unit 222 IFFT conversion unit 223 Removal unit 224 Comparison unit 225 Coefficient unit 23 Operation mechanism 24 Switching mechanism 3 Sound input device 30 Sound input mechanism 4 sound processing device 40 input mechanism 41 recognition mechanism 42 control mechanism 5 moving body 100 echo suppression program

  Hereinafter, the present invention will be described in detail with reference to the drawings illustrating embodiments thereof.

Embodiment 1 FIG.
FIG. 4 is a block diagram schematically showing a configuration example of the echo suppression system according to Embodiment 1 of the present invention. In FIG. 4, reference numeral 1 denotes an echo suppression device used in an echo suppression system applied to a system such as a car navigation system. The echo suppression device 1 includes a plurality of sound output mechanisms such as speakers that output a plurality of channels. In cooperation with a sound output device 2 such as multi-channel audio having 20, 20,..., A sound input device 3 having a sound input mechanism 30 using a microphone such as a condenser microphone, and a sound processing device 4 such as a speech recognition system. Operate. The echo suppression device 1, the sound output device 2, the sound input device 3, and the sound processing device 4 are mounted on a moving body 5 such as a vehicle. The echo suppression system of the present invention is not limited to the car navigation system illustrated in FIG. 4 but can be applied to various systems such as an audio system and a video conference system.

  The sound output device 2 reads sound data from a recording medium such as a music CD (Compact Disc) or DVD (Digital Versatile Disc) for recording sound data, and generates sound signals for a plurality of channels based on the read sound data. The sound signal generating mechanism 21 outputs a plurality of generated sound signals as a multi-channel audio signal, and a DSP (Digital Signal Processor) that allows components of different frequency bands to pass through the sound signals for a plurality of channels. The sound signals for a plurality of channels that have passed through the passage mechanism 22 are output from the sound output mechanisms 20, 20,. The sound output device 2 transmits sound signals for a plurality of channels that have passed through the passage mechanism 22 to the echo suppression device 1. Note that the sound signal output from the sound signal generation mechanism 21 is an analog signal, and is converted to a digital signal by an A / D (Analog to Digital) converter (not shown) and then output to the passage mechanism 22. Further, a sound signal which is a digital signal is output from the passage mechanism 22, converted into an analog signal by a D / A (Digital to Analog) converter (not shown), and then to the sound output mechanism 20, 20... And the echo suppressor 1. Output. The sound output mechanisms 20, 20,... Are each assigned a channel, and each sound signal that has passed through the passage mechanism 22 is output to the sound output mechanisms 20, 20,. In the echo suppression system of the present invention, the sound output mechanisms 20, 20,... Have a function of outputting sound based on the sound signal which is a digital signal, and all inputs and outputs between the mechanisms and the apparatus are performed by digital signals. It is possible to design appropriately.

  The sound input device 3 uses the sound input mechanism 30 to generate a sound signal that is an analog signal based on the input sound, amplifies the generated sound signal with an amplifier (not shown) such as a gain amplifier, and the amplified sound signal Is sampled at a sampling frequency such as 8000 Hz and 12000 Hz by an A / D converter (not shown) and converted into a digital signal, and the sound signal converted into the digital signal is output to the echo suppressor 1.

  The echo suppression apparatus 1 includes a first input mechanism 10 to which sound signals for a plurality of channels output from the sound output apparatus 2 are input, and a sound signal output from the sound input apparatus 3 as an observation sound signal. A two-input mechanism 11, a suppression mechanism 12 such as a DSP that suppresses an echo component included in the observation sound signal, and an output mechanism 13 that outputs the observation sound signal in which the echo component is suppressed by the suppression mechanism 12 to the sound processing device 4. It has.

  The suppression mechanism 12 incorporates an echo suppression program 100 and firmware such as data. By executing the echo suppression program 100 of the present invention that is incorporated as firmware, a plurality of inputs input by the first input mechanism 10 are performed. An adder 120 that adds the sound signals for the channels to generate a reference sound signal, and a correction unit 121 that corrects the observation sound signal to suppress echoes included in the observation sound signal based on the observation sound signal and the reference sound signal. Various program modules such as are generated. Then, by executing the generated various program modules, the suppression mechanism 12 included in the echo suppression device 1 inputs echoes based on sounds output from the sound output mechanisms 20, 20,... To the sound input mechanism 30. It functions as an echo canceller that removes sound.

  Further, the first input mechanism 10 includes an A / D converter that converts sound signals for a plurality of channels, which are analog signals input from the sound output device 2, into digital signals. However, when a sound signal is input as a digital signal from the first input mechanism 10, an A / D converter is not necessary. Further, the functions of the amplifier and the A / D converter included in the sound input device 3 may be provided in the second input mechanism 11.

  The configuration example shown in FIG. 4 is merely an example of an infinite form of the present invention, and the hardware and software configurations can be changed as necessary. For example, each program module can also be configured as hardware using an arithmetic circuit such as a VLSI, and the implementation form can be changed as appropriate, for example, by providing the sound output device 2 with the function of the adder 120. Can do. Further, when the adder 120 is configured as hardware outside the suppression mechanism 12, the adder 120 may be configured using the adder 120 using a mixing circuit that mixes analog signals. The sound signals of a plurality of channels that are input analog signals are mixed (added) by the adding unit 120 and then converted into digital signals by an A / D converter (not shown).

  4, 1ch,..., Nch (n is a natural number) indicate a plurality of channels, and x1 (t),..., Xn (t) are output from the sound output mechanism 20 to the passage mechanism 22. The sound signals from channel to n channel are shown. The variable t is a sample number for identifying each sample when a sound signal, which is an analog signal, is sampled at a sampling frequency such as 8000 Hz or 12000 Hz and converted into a digital signal. Further, x1_f (t),..., Xn_f (t) indicate sound signals from the 1 channel to the n channel that have passed through the passing mechanism 22, and the sound signals x1_f (t),. By adding (t), the reference sound signal x_f (t) is generated. Further, y (t) indicates an observation sound signal, and r (t) is a sound signal indicating a suppression result in which the echo component of the observation sound signal y (t) is suppressed by the suppression mechanism 12. The result r (t) is output to the sound processing device 4, and the sound processing device 4 performs processing such as speech recognition.

  FIG. 5 is a functional block diagram showing a configuration example of the passage mechanism 22 of the sound output device 2 according to Embodiment 1 of the present invention. The passing mechanism 22 using the DSP converts the input sound signals of a plurality of channels into sound signals of components on the frequency axis by FFT (Fast Fourier Transformation) processing, respectively. ..., a plurality of band-pass filter sections 221, 221 that allow components of different frequency bands to pass through a plurality of sound signals converted into components on the frequency axis, and frequency axes that pass components of the respective frequency bands Various program modules such as a plurality of IFFT converters 222, 222,... That convert a plurality of sound signals converted into the above components into sound signals on the time axis by IFFT (inverse Fourier transform) processing are executed. To do. When converting the sound signal, the FFT conversion unit 220 generates a sound signal in units of frames with, for example, a signal of 512 samples as one frame. Each frame overlaps by about 128 to 256 samples, and each frame is generally used in the field of speech recognition such as a Hamming window, a window function such as a Hanning window, and filtering using a high frequency enhancement filter. Frame processing is performed. Then, the FFT conversion unit 220 performs FFT processing on the generated sound signal in units of frames. The band pass filter units 221, 221,... Each correspond to a channel, and each sound signal converted into a component on the frequency axis passes through the band pass filter units 221, 221,. To do. Further, each of the band-pass filter units 221, 221... Is preset with different filter coefficients indicating the transmissivity for each frequency, and performs filtering on the sound signal based on the set filter coefficients.

  5, x1 (t),..., Xn (t) are sound signals of a plurality of channels input to the FFT converters 220, 220,..., And the FFT converters 220, 220,. , Xn (t) are converted to components on the frequency axis and a plurality of sound signals X1 (f),..., Xn (f) are output to the bandpass filter units 221, 221. To do. The variable f indicates a frequency. Further, the band pass filter units 221, 221,... Output a plurality of sound signals X1_f (f),..., Xn_f (f) through which the components of the respective frequency bands are passed to the IFFT conversion units 222, 222,. Further, the IFFT converters 222, 222,... Respectively convert a plurality of sound signals X1_f (f),..., Xn_f (f) through which the components of the respective frequency bands have passed into sound signals x1_f ( t),..., xn_f (t).

  FIG. 6 is a graph showing the filter coefficient of the band pass filter unit 221 of the pass mechanism 22 provided in the sound output device 2 according to Embodiment 1 of the present invention. FIGS. 6A, 6B, and 6C show the sound signals X1 (f), X2 (f), and Xn (f) of the first channel 1ch, the second channel 2ch, and the nth channel nch, respectively. The coefficient filters C1 (f), C2 (f), and Cn (f) are shown, and the horizontal axis is the frequency f and the vertical axis is the filter coefficient. The filter coefficient is a coefficient set for each frequency multiplied by the sound signal, and each of the band pass filter units 221, 221,... Has converted the set filter coefficient into a component on the frequency axis. It is a program module that performs processing for multiplying sound signals. By the processing of the band pass filter unit 221, the component of the frequency band whose filter coefficient is 1.0 passes through the band pass filter unit 221, but the component of the frequency band whose filter coefficient is 0.0 is When passing through the band pass filter unit 221, the amplitude becomes 0 and is removed by the band pass filter unit 221. As shown in FIGS. 6A, 6B, and 6C, the filter coefficients of the band pass filter units 221, 221,... Are set so as to pass components in different frequency bands.

  In each of the band pass filter units 221, 221,..., The frequency band is divided at equal intervals, and the filter coefficients are set so that the components of the divided frequency bands pass through only one of the band pass filter units 221. ing. By setting the filter coefficients as shown in FIG. 6, each bandpass filter unit 221, 221... Functions as a comb filter. When setting the filter coefficients of the band pass filter units 221, 221,..., It is divided so that the logarithmic values of the frequencies are equally spaced, and each of the divided frequency band components is only one of the band pass filter units 221. A comb filter may be formed by setting a filter coefficient so as to pass, and by dividing a frequency band using a logarithmic value, it is possible to output a sound with a less uncomfortable feeling during listening.

  FIG. 7 is a functional block diagram showing a configuration example of the suppression mechanism 12 included in the echo suppression device 1 according to Embodiment 1 of the present invention. The suppression mechanism 12 using a DSP executes program modules such as the addition unit 120 and the correction unit 121 as described above.

  Further, the correction unit 121 filters the reference sound signal x_f (t) by a several hundredth-order product-sum operation using a filter coefficient set for each frequency required for echo estimation, so that the observed sound signal y As a correction amount required for correcting (t), a linear FIR filter (correction filter) unit 1210 for deriving an echo signal x ′ (t), and subtracting the echo signal x ′ (t) from the observed sound signal y (t) Thus, the observation sound signal y (t) is corrected and the observation sound signal in which the echo component is suppressed is output as the suppression result r (t), and the corrected observation sound signal, that is, the suppression result r (t t), the filter coefficient updating unit 1212 that calculates and updates the filter coefficient of the linear FIR filter unit 1210 by adaptive processing using the learning identification method, and the speech based on the suppression result r (t). There executes a detection unit 1213 for detecting whether a state of single-talk not speaking or the state of double-talk which is uttered as a submodule. Then, the detection unit 1213 detects the single talk state and the double talk state based on the intensity change of the suppression result r (t). When the double talk state is detected, the filter coefficient update unit 1212 calculates and updates the filter coefficient. Stop. In the processing of the correction unit 121, the reference sound signal x_f (t) and the observation sound signal y (t) are subjected to framing and FFT processing for converting into components on the frequency axis as necessary. It is also possible to configure so that a suppression result r (t) that has been subjected to echo suppression processing and further subjected to IFFT processing for conversion to a component on the time axis is output.

  FIG. 8 is a block diagram showing a configuration example of the sound processing device 4 according to Embodiment 1 of the present invention. The sound processing device 4 receives the suppression result r (t) output from the output mechanism 13 of the echo suppression device 1 and the voice based on the suppression result r (t) input from the input mechanism 40. A recognition mechanism 41 that performs recognition processing and a control mechanism 42 that controls navigation processing based on the recognition result of the recognition mechanism 41 are provided. The control by the control mechanism 42 is processing inherent in the car navigation system such as input of a destination of the moving body 5, derivation of a travel route, and display of a travel route based on a command recognized from a voice uttered by a speaker. When the system of the present invention is applied to an audio system, the sound processing device 4 functions as a control device that controls the echo suppression device 1, the sound output device 2, and the sound input device 3, and a voice command generated by a speaker. For example, the control mechanism 42 controls the echo suppression device 1, the sound output device 2, and the sound input device 3 based on commands such as sound source switching, reproduction start, and voice recognition end.

  Next, processing of each device provided in the echo suppression system according to Embodiment 1 of the present invention will be described. FIG. 9 is a flowchart showing an example of sound output processing of the sound output device 2 according to Embodiment 1 of the present invention. The sound output device 2 generates sound signals x1 (t),..., Xn (t) for a plurality of channels by the sound signal generation mechanism 21 (S101), and the generated sound signals x1 (t),. t) is output to the passing mechanism 22.

  The passage mechanism 22 of the sound output device 2 converts the input sound signals x1 (t),..., Xn (t) of a plurality of channels into digital signals, framing them, and using the FFT converters 220, 220,. The sound signals are converted into sound signals X1 (f),..., Xn (f) on the frequency axis by FFT processing (S102), and the sound signals X1 ( f),..., Xn (f) are passed to the band-pass filter units 221, 221. In step S102, as a method of converting to a component on the frequency axis, it is not always necessary to use FFT, and other conversion methods such as DCT (Discrete Cosine Transform) may be used.

  The pass mechanism 22 of the sound output device 2 respectively receives a plurality of sound signals X1 (f),..., Xn (f) converted into components on the frequency axis by the band pass filter units 221, 221. The components X1_f (f),..., Xn_f (f) of different frequency bands are passed (S103), and the IFFT converters 222, 222,. , Xn_f (t) (S104), converted into an analog signal, output to the sound output mechanism 20, 20,..., And output to the echo suppressor 1 (S105).

  The sound output mechanisms 20, 20,... Of the sound output device 2 output sounds based on the input sound signals x1_f (t),..., Xn_f (t) (S106).

  FIG. 10 is a flowchart showing an example of sound input processing of the sound input device 3 according to Embodiment 1 of the present invention. The sound input device 3 inputs a sound by the sound input mechanism 30 (S201), generates a sound signal based on the input sound, and uses the generated sound signal as an observed sound signal y (t) as an echo suppression device 1 (S202). In step S201, the sound input device 3 receives sound from a sound field from which the sound output device 2 outputs sound. Therefore, the sound input from the sound input device 3 may be mixed with the sound output from the sound output device 2.

  FIG. 11 is a flowchart showing an example of echo suppression processing of the echo suppression apparatus 1 according to Embodiment 1 of the present invention. The echo suppression device 1 inputs sound signals x1_f (t),..., Xn_f (t) for a plurality of channels output from the sound output device 2 by the first input mechanism 10 (S301), and the input sound signals x1_f (t),..., xn_f (t) are converted into digital signals and transferred to the suppression mechanism 12, and the observation sound signal y (t) output from the sound input device 3 is input by the second input mechanism 11. (S302), the input observation sound signal y (t) is converted into a digital signal and passed to the suppression mechanism 12. Note that the processes in steps S301 and S302 are performed substantially in parallel.

  The suppression mechanism 12 of the echo suppressor 1 adds the sound signals x1_f (t),..., Xn_f (t) for a plurality of channels output from the sound output device 2 by the adding unit 120 to add the reference sound signal x_f (x ) Is generated (S303), and the generated reference sound signal x_f (x) is passed to the correction unit 121.

  The suppression mechanism 12 of the echo suppression apparatus 1 filters the reference sound signal x_f (x) by the product-sum operation using the set filter coefficient in the linear FIR filter unit 1210 as processing of the correction unit 121. Thus, the echo signal x ′ (t) is derived as a correction amount required for correcting the observed sound signal y (t) (S304), and the echo signal x ′ (t), which is the derived correction amount, is passed to the subtracting unit 1211.

  The suppression mechanism 12 of the echo suppressor 1 subtracts the echo signal x ′ (t) from the observation sound signal y (t) by the subtraction unit 1211 as the processing of the correction unit 121, thereby obtaining the observation sound signal y (t). It correct | amends (S305), converts the suppression result r (t) which is the corrected observation sound signal y (t) into an analog signal, and outputs the converted suppression result r (t) to the sound processor 4 (S306). The sound processing device 4 executes processing such as speech recognition based on the input suppression result r (t).

  FIG. 12 is a flowchart showing an example of the filter coefficient update process of the echo suppression apparatus 1 according to Embodiment 1 of the present invention. In the echo suppression apparatus 1, in parallel with the echo suppression process using FIG. 11, the update process of the filter coefficient used for the derivation of the echo signal is executed. The suppression mechanism 12 of the echo suppressor 1 is a double-talk state in which the speaker is speaking based on the suppression result r (t) output from the subtraction unit 1211 by the detection unit 1213 as the processing of the correction unit 121. It is detected whether the single talk is not uttered (S401), and the detection result indicating the detected result is passed to the filter coefficient updating unit 1212.

  The suppression mechanism 12 of the echo suppression apparatus 1 determines whether or not it is necessary to calculate and update the filter coefficient based on the detection result received from the detection unit 1213 by the filter coefficient update unit 1212 as processing of the correction unit 121 (S402). ). In step S402, when the detection result indicates single talk, calculation and update of the filter coefficient are executed, and when double detection is indicated, calculation and update of the filter coefficient are stopped.

  If it is determined in step S402 that the filter coefficient needs to be calculated and updated (S402: YES), the suppression mechanism 12 of the echo suppression device 1 performs processing of the correction unit 121 by the filter coefficient update unit 1212 from the subtraction unit 1211. Based on the output suppression result r (t), filter coefficients are calculated by adaptive processing using a learning identification method (S403), and the filter coefficients used in the linear FIR filter unit 1210 are set in step S403. (S404), the process returns to step S401, and the subsequent processing is repeated.

  When it is determined in step S402 that calculation and update of the filter coefficient are not required (S402: NO), the suppression mechanism 12 of the echo suppression device 1 calculates the filter coefficient by the filter coefficient update unit 1212 as processing of the correction unit 121. And without updating, it returns to step S401 and repeats the subsequent processing.

  When sounds of a plurality of channels are output from the sound output device 2 to the sound field, the sound signal input from the sound field includes an echo based on the sound output from the sound output device 2. When the sound field maintains linearity with respect to sound, the sound output from the sound output device 2 is input to the sound input device 3 while maintaining its frequency. In addition, the reference sound signal x_f (t) is a signal obtained by adding the sound signals x1_f (t),..., Xn_f (t) of a plurality of channels to which different frequency bands are assigned, and therefore the reference sound at any frequency f The relationship between the signal x_f (t) and the observation sound signal y (t) can be regarded as equivalent to the relationship in the case of outputting a sound of one channel. Therefore, in the echo suppression system of the present application, it is possible to suppress echoes with the same accuracy as when outputting a sound signal of one channel even when outputting a sound signal of a plurality of channels.

Embodiment 2. FIG.
The second embodiment is a form in which efficient filtering is realized for the sound signals of a plurality of channels realized by using the processed sound signal obtained by processing the original sound signal by a predetermined method in the first embodiment. Here, as the second embodiment, the first sound signal of the L (Left) channel, the second sound signal of the R (Right) channel, the chord of the C (Center) channel that is the sum of the first sound signal and the second sound signal. Signal, a delayed first sound signal of an sL (Surround Left) channel obtained by delaying the first sound signal by a predetermined time, and a pseudo second sound signal of an sR (Surround Right) channel obtained by delaying the second sound signal by a predetermined time. An example applied to a 5-channel stereo system will be described. In the pseudo 5 channel, the first sound signal and the second sound signal are original sound signals, and the chord signal, the delayed first sound signal and the delayed second sound signal are processed sound signals processed by adding or delaying the original sound signals. is there. However, the echo suppression system according to the second embodiment of the present invention is not necessarily limited to the pseudo 5 channel, but the pseudo 4 channel including the first sound signal, the second sound signal, the delayed first sound signal, and the delayed second sound signal. The present invention can be developed in various forms such as a form applied to the stereo system and a form using a delayed chord signal obtained by delaying the chord signal as a pseudo signal.

  In the following description, components similar to those in the first embodiment are denoted by the same reference numerals as those in the first embodiment, and detailed description thereof is omitted. Since the configuration example of the echo suppression system in the second embodiment is the same as that in the first embodiment shown in FIG. 4, the first embodiment is referred to and the description thereof is omitted.

  FIG. 13 is a functional block diagram showing a configuration example of the passing mechanism 22 of the sound output device 2 according to Embodiment 2 of the present invention. The configuration of the passage mechanism 22 of the sound output device 2 according to the second embodiment is substantially the same as that of the first embodiment. In FIG. 13, xL (t), xR (t), xC (t), xsL (t), xsR (t) are the first sound signal input to the FFT converters 220, 220,. A sound signal, a chord signal, a delayed first sound signal, and a delayed second sound signal are shown. These sound signals xL (t), xR (t), xC (t), xsL (t), xsR (t) are converted into components on the frequency axis by the FFT converters 220, 220,. Are XL (f), XR (f), XC (f), XsL (f), and XsR (f). Further, these sound signals XL (f), XR (f), XC (f), XsL (f), XsR (f) converted into components on the frequency axis pass through the band-pass filter units 221, 221. The sound signals obtained are XL_f (f), XR_f (f), XC_f (f), XsL_f (f), and XsR_f (f). The sound signals on the time axis obtained by converting these sound signals XL_f (f), XR_f (f), XC_f (f), XsL_f (f), XsR_f (f) by the IFFT converters 222, 222,. (T), xR_f (t), xC_f (t), xsL_f (t), xsR_f (t).

  FIG. 14 is a graph showing filter coefficients of the band pass filter unit 221 of the pass mechanism 22 provided in the sound output device 2 according to Embodiment 2 of the present invention. FIGS. 14 (a), (b), (c), (d), and (e) show the L channel first sound signal XL (f), the R channel second sound signal XR (f), and the C channel, respectively. Coefficient filters CL (f), CR (f), CC (f) for the chord signal XC (f), the delayed first sound signal XsL (f) of the sL channel and the delayed second sound signal XsR (f) of the sR channel , CsL (f) and CsR (f), the horizontal axis is the frequency f, and the vertical axis is the filter coefficient.

  The coefficient filter CL (f) for the first sound signal XL (f) shown in FIG. 14A and the coefficient filter CR (f) for the second sound signal XR (f) shown in FIG. It is a filter set to pass components of different frequency bands. Since the chord signal XC (f) is a sound signal obtained by adding the first sound signal XL (f) and the second sound signal XR (f), the chord signal XC (f) has a first sound signal XL. The coefficient filter CC (f) shown in FIG. 14C, which is the sum of the coefficient filter CL (f) for (f) and the coefficient filter CR (f) for the second sound signal XR (f), is used. In the example shown as the second embodiment, the coefficient filter CC (f) is a filter that passes the components of the entire frequency band to be filtered as shown in FIG. 14C, and the filter coefficient regardless of the frequency. Becomes 1.0. Since the delayed first sound signal XsL (f) is a sound signal obtained by delaying the first sound signal XL (f), the first sound signal XL (f The coefficient filter CsL (f) shown in FIG. 14 (d) obtained by diverting (copying) the coefficient filter CL (f) to () is used. Therefore, the coefficient filter CL (f) for the first sound signal XL (f) and the coefficient filter CsL (f) for the delayed first sound signal XsL (f) are the same filter. Further, since the delayed second sound signal XsR (f) is a sound signal obtained by delaying the second sound signal XR (f), the second sound signal XR (f) ) Is used (copied) as a coefficient filter CsR (f) shown in FIG. 14E. Accordingly, the coefficient filter CR (f) for the second sound signal XR (f) and the coefficient filter CsR (f) for the delayed second sound signal XsR (f) are filters having the same filter coefficient.

  As described above, in the second embodiment, the bandpass filter units 221, 221... Are configured by a memory for storing signals for five channels and a program module for performing filter processing in order to process signals for five channels. The As the filter coefficients used in the respective program modules, filter coefficients uniquely set for the first sound signal XL (f) and the second sound signal XR (f) are used. For the chord signal XC (f), a filter coefficient calculated as the sum of the filter coefficient of the first sound signal XL (f) and the filter coefficient of the second sound signal XR (f) is used. Further, for the delayed first sound signal XsL (f), a filter coefficient obtained by copying the filter coefficient of the first sound signal XL (f) is used, and for the delayed second sound signal XsR (f), the first A filter coefficient obtained by copying the filter coefficient of the two-tone signal XR (f) is used. The program module that performs the filtering process for each channel may use one program module in a time-sharing manner and reset the filter coefficient for each signal.

  Since other configurations and processes are the same as those in the first embodiment, the first embodiment will be referred to and description thereof will be omitted.

Embodiment 3 FIG.
The third embodiment corresponds to a preset removal band such as a frequency band corresponding to a voice uttered by a person with respect to sound signals for a plurality of channels generated by the sound signal generation mechanism in the first embodiment. In this configuration, the frequency band component is not allowed to pass through any sound signal, and double talk and single talk are detected based on the frequency band component corresponding to the removal band.

  In the following description, components similar to those in the first embodiment are denoted by the same reference numerals as those in the first embodiment, and detailed description thereof is omitted. Since the configuration example of the echo suppression system in the third embodiment is the same as that in the first embodiment shown in FIG. 4, the first embodiment is referred to and the description thereof is omitted.

  FIG. 15 is a functional block diagram showing a configuration example of the passage mechanism 22 of the sound output device 2 according to Embodiment 3 of the present invention. The passing mechanism 22 according to Embodiment 3 includes removal units 223, 223,... That remove components in a frequency band corresponding to a preset removal band f_cut. The removal units 223, 223,... Are removed from a plurality of sound signals X1_f (f),..., Xn_f (f) converted into components on the frequency axis that have passed through the bandpass filter units 221, 221. The frequency band component corresponding to f_cut is removed, and signals X1_f_c (f),..., Xn_f_c (f) from which the removed band f_cut components are removed are output to IFFT converters 222, 222,. .., Xn_f_c (f) are passed through the removal units 223, 223,... To the sound signal x1_f_c (on the time axis by IFFT processing, respectively. t),..., xn_f_c (t). The IFFT converters 222, 222,... Output the sound signals x1_f_c (t),..., Xn_f_c (t) to the sound output mechanisms 20, 20,.

  FIG. 16 is a graph showing filter coefficients of the removal unit 223 of the passage mechanism 22 included in the sound output device 2 according to Embodiment 3 of the present invention. FIG. 16 shows a removal coefficient filter CCUT (f), which is a graph showing the relationship with the horizontal axis representing frequency and the vertical axis representing filter coefficient. As shown in FIG. 16, in the removal coefficient filter CCUT (f), the filter coefficient of the removal band f_cut corresponding to human speech is 0.0, and 1.0 for other bands. In this way, the sound output device 2 according to Embodiment 3 sets the removal band f_cut, thereby removing the frequency component corresponding to the removal band f_cut from the sound of all the channels to be output. Instead of providing the removal unit 223, the band-pass filter units 221, 221,... Set the filter coefficients of the band-pass filter units 221, 221,... So that the frequency components corresponding to the removal band f_cut are blocked. You may make it do.

  FIG. 17 is a functional block diagram showing a configuration example of the suppression mechanism 12 included in the echo suppression apparatus 1 according to Embodiment 3 of the present invention. The correction unit 121 of the suppression mechanism 12 according to Embodiment 3 executes, as a submodule, a removal band pass filter unit 1214 that passes only a frequency component corresponding to the removal band f_cut from the observed sound signal y (t). The detection unit 1213 detects whether the signal is in a single talk state or a double talk state based on the detection sound signal y_p (t) for detection that has passed through the removal band pass filter unit 1214. When the sound field in which the sound output device 2 outputs sound of a plurality of channels maintains linearity with respect to the sound, the removal band f_cut of the observed sound signal y (t) includes an echo component based on the output sound. It is not a habit. Therefore, when the single talk state and the double talk state are detected based on the detection sound signal y_p (t) for detection in which only the component of the removal band f_cut is passed, the suppression result as shown in the first embodiment. Compared to the case of detecting from r (t), it is possible to detect the state with higher accuracy.

  FIG. 18 is a graph showing filter coefficients of the removal band-pass filter unit 1214 of the suppression mechanism 12 included in the echo suppression apparatus 1 according to Embodiment 3 of the present invention. FIG. 18 shows the removal band pass coefficient filter CPASS (f) set in the removal band pass filter unit 1214, and is a graph showing the relationship with the horizontal axis representing frequency and the vertical axis representing filter coefficient. As shown in FIG. 18, in the removal band pass coefficient filter CPASS (f), the filter coefficient of the removal band f_cut is 1.0, and 0.0 for the other bands. In this way, the removal band pass filter unit 1214 according to Embodiment 3 passes only the frequency component corresponding to the removal band f_cut from the observation sound signal y (t).

  Next, processing of each device provided in the echo suppression system according to Embodiment 3 of the present invention will be described. FIG. 19 is a flowchart showing an example of sound output processing of the sound output device 2 according to Embodiment 3 of the present invention. In the third embodiment, the frequency components corresponding to the removal band f_cut are removed by the removal unit 223 of the passing mechanism 22 after performing the processing of steps S101 to S103 shown in the flowchart of the sound output processing according to the first embodiment. Execute the process.

  The sound output device 2 executes the processes of steps S101 to S103 shown in the first embodiment. Then, the passing mechanism 22 of the sound output device 2 includes a plurality of sound signals X1_f (f),..., Xn_f converted by the removing unit 223 into components on the frequency axis that have passed through the bandpass filter units 221, 221. From (f), the components of the frequency band corresponding to the removal band f_cut are removed (S501), and the signals X1_f_c (f),. Output to…. Then, the sound output device 2 executes the processes after step S104 shown in the sound output process of the first embodiment.

  Since the sound input process of the sound input device 3 and the echo suppression process of the echo suppression device 1 according to the third embodiment are the same as those in the first embodiment, the first embodiment is referred to and the description thereof is omitted. . However, in the echo suppression processing, the addition unit 120 of the suppression mechanism 12 adds the sound signals x1_f_c (t),..., Xn_f_c (t) for a plurality of channels from which the components of the removal band f_cut are removed to add the reference sound signal x_f ′ (X) is generated.

  FIG. 20 is a flowchart showing an example of filter coefficient update processing of the echo suppression apparatus 1 according to Embodiment 3 of the present invention. In the suppression mechanism 12 of the echo suppression device 1, as processing of the correction unit 121, a speaker speaks based on the detection sound signal y_p (t) for detection that has passed through the removal band pass filter unit 1214 by the detection unit 1213. It is detected whether it is a double talk state or a single talk state that is not uttered (S601), and a detection result indicating the detection result is passed to the filter coefficient updating unit 1212. Then, the echo suppression apparatus 1 executes the processes after step S402 shown in the filter coefficient update process of the first embodiment.

Embodiment 4 FIG.
The fourth embodiment is a form in which the echo suppression method of the present invention is executed only when a predetermined operation such as pressing a speech switch of the car navigation system is performed in the first embodiment. In the following description, components similar to those in the first embodiment are denoted by the same reference numerals as those in the first embodiment, and detailed description thereof is omitted.

  FIG. 21 is a block diagram schematically showing a configuration example of an echo suppression system according to Embodiment 4 of the present invention. The sound output device 2 according to the fourth embodiment includes an operation mechanism 23 that functions as an utterance switch, and switching mechanisms 24, 24,... Such as a switch that performs a signal path switching process based on the operation of the operation mechanism 23. . The switching mechanisms 24, 24,... Are respectively disposed on the signal lines for transmitting a plurality of channels of sound signals from the sound signal generation mechanism 21 to the passage mechanism 22, and the plurality of channels output from the sound signal generation mechanism 21. The sound signal is output to the passage mechanism 22 or the plurality of sound output mechanisms 20, 20,.

  When the speaker performs an operation such as pressing the operation mechanism 23, the operation mechanism 23 outputs an operation signal to the switching mechanisms 24, 24,..., And the switching mechanisms 24, 24,. A signal path switching process is performed so that a plurality of channels of sound signals output from the sound signal generating mechanism 21 are output to the passing mechanism 22 for a predetermined time from the input of the operation signal. While the sound signals of a plurality of channels are being output to the passing mechanism 22, the echo suppression system of the present invention performs various processes described in the first embodiment.

  When the speaker does not operate the operation mechanism 23, or when a certain time has elapsed since the operation, the switching mechanisms 24, 24, ... output sound signals of a plurality of channels output from the sound signal generation mechanism 21 as sound outputs. A signal path is set so as to output to the mechanisms 20, 20,.

  As described above, in the fourth embodiment, the echo suppression method of the present invention is executed only when the speaker performs an operation, and in a normal time when the operation is not performed, a plurality of channels output from the sound signal generating mechanism 21 are output. Are output as they are to the sound output mechanisms 20, 20,..., So that the frequency components of the sound signals are not selectively passed, so that the sound quality of the output sound can be maintained.

  Since other configurations and processes are the same as those in the first embodiment, the first embodiment will be referred to and description thereof will be omitted.

Embodiment 5 FIG.
The fifth embodiment is a form in which the processing by the passage mechanism is dynamically changed based on the correlation between sound signals in the first embodiment.

  In the following description, components similar to those in the first embodiment are denoted by the same reference numerals as those in the first embodiment, and detailed description thereof is omitted. Since the configuration example of the echo suppression system in the fifth embodiment is the same as that in the first embodiment shown in FIG. 4, the first embodiment is referred to and the description thereof is omitted.

  FIG. 22 is a functional block diagram showing a configuration example of the passing mechanism 22 of the sound output device 2 according to Embodiment 5 of the present invention. The passing mechanism 22 according to the fifth embodiment compares the amplitudes of the sound signals X1 (f),..., Xn (f) of the respective channels converted into components on the frequency axis in units of frames, and compares the compared amplitudes. Is provided with a comparison unit 224 that dynamically changes the filter coefficients of the bandpass filter units 221, 221. Instead of comparing the amplitudes of the sound signals X1 (f),..., Xn (f), the power that is the square of the amplitude may be compared as an alternative value of the amplitude.

  The comparison unit 224 derives a correlation between the sound signals X1 (f),..., Xn (f) for each frequency band at a predetermined number of frames or at predetermined time intervals. The correlation is derived by comparing the amplitudes of the sound signals. If it is determined by comparison that there is one sound signal having an amplitude greater than a predetermined value relative to the amplitude of all other sound signals, the correlation between the sound signals X1 (f),..., Xn (f). Is determined to be weak, and in that case, one sound signal having the largest amplitude is derived. The comparison of the amplitude of one sound signal with another sound signal is, for example, the ratio of the amplitude of one sound signal to the amplitude of the other sound signal, or the difference obtained by subtracting the amplitude of the other sound signal from the amplitude of the one sound signal. By determining whether or not the comparison result compared by a predetermined method such as a ratio or difference of amplitude absolute values or a power ratio or difference which is the square of amplitude is greater than or equal to a predetermined value set in advance Done. The predetermined value used for the comparison is set so as to satisfy a condition that it can be said that the amplitude of all other sound signals is small enough to be ignored with respect to the amplitude of one sound signal.

  For example, when the amplitude of the n-th channel nch at the frequency f is represented by nch (f) and the predetermined value is α, each sound signal X1 (f),..., Xn (f ) Is weak and the sound signal having the largest amplitude is determined to be the sound signal Xn (f) of the nth channel nch.

  nch (f) / 1ch (f) ≧ α, nch (f) / 2ch (f) ≧ α, Formula (1)

  The comparison unit 224 derives filter coefficients of the band-pass filter units 221, 221,... At the deflection frequency fd indicating the frequency f determined that the correlation between the sound signals X1 (f),..., Xn (f) is weak. To do. The comparison unit 224 derives 1.0 as the filter coefficient of the band pass filter unit 221 related to the sound signal of the channel having the maximum amplitude, and 0 as the filter coefficient of the band pass filter unit 221 related to the sound signal of the other channel. .0 is derived. The comparison unit 224 passes the filter coefficients of the band pass filter units 221, 221,... At the deflection frequency fd to the corresponding band pass filter units 221, 221,.

  When the band pass filter units 221, 221,... Accept filter coefficients at the deflection frequency fd, the filter coefficients at the deflection frequency fd are changed to the accepted filter coefficients. That is, it is set so that only the sound signal having the maximum amplitude is passed.

  When it is determined by comparison of the comparison unit 224 that there is no one sound signal having an amplitude greater than a predetermined value with respect to the amplitudes of all other sound signals, the sound signals X1 (f),. It is judged that the correlation between f) is strong. When determining that the correlation is strong, the comparison unit 224 does not derive the deflection frequency fd and the filter coefficient. Therefore, the band pass filter units 221, 221,... Execute the same processing as in the first embodiment.

  A situation in which the correlation between the sound signals X1 (f),..., Xn (f) is low and the amplitude related to one sound signal is larger than the amplitude related to the other sound signal is, for example, a multi-channel orchestra performance. Occurs during a solo performance in which the output is biased to one of the channels during playback. In such a situation, if the sound signal related to the instrument performing the solo performance is suppressed by the band-pass filter units 221, 221,..., The listener will feel uncomfortable with the output sound. In the fifth embodiment, in such a situation, the sound based on the sound signal related to the musical instrument performing the solo performance is selected as the output sound, so that it is possible to output a sound with less sense of incongruity.

  Next, processing of each device provided in the echo suppression system according to Embodiment 5 of the present invention will be described. FIG. 23 is a flowchart showing an example of sound output processing of the sound output device 2 according to Embodiment 5 of the present invention. In the fifth embodiment, the processes of steps S101 to S102 of the sound output process according to the first embodiment are executed. Then, the passing mechanism 22 of the sound output device 2 uses the comparison unit 224 to compare the amplitudes of the sound signals X1 (f),..., Xn (f) for each frequency band, and each sound signal X1 (f). ,..., Xn (f) are weak, and it is determined whether or not there is one sound signal having an amplitude greater than a predetermined value with respect to the amplitude of all other sound signals (S701).

  In step S701, it is determined that the correlation between the sound signals X1 (f),..., Xn (f) is weak, and there is one sound signal having an amplitude greater than a predetermined value with respect to the amplitudes of all other sound signals. In the case (S701: YES), the passage mechanism 22 of the sound output device 2 uses the comparison unit 224 to identify one sound signal having the largest amplitude at the frequency f, that is, the deflection frequency fd, and use the identified sound signal. Are derived as the filter coefficients at the deflection frequency fd to the corresponding band-pass filter sections 221, 221,..., Respectively. In step S702, the filter coefficient of the band-pass filter unit 221 relating to one sound signal having the largest amplitude is derived as 1.0, and the filter coefficient of the band-pass filter units 221, 221,. Derived as .0.

  The passing mechanism 22 of the sound output device 2 changes the setting of the filter coefficient at the deflection frequency fd of each of the band-pass filter units 221, 221... To the accepted filter coefficient (S703). Then, the sound output device 2 executes the processes after step S103 shown in the sound output process of the first embodiment.

  In step S701, if there is a strong correlation between the sound signals X1 (f),..., Xn (f) and there is no sound signal having an amplitude greater than a predetermined value with respect to the amplitudes of all other sound signals. When it determines (S701: NO), the passage mechanism 22 of the sound output device 2 cancels the change of the filter coefficient setting of each bandpass filter unit 221,221,... (S704). Canceling the change in step S704 is a process of canceling the changed setting and returning it to the original filter coefficient setting when the filter coefficient setting has been changed in step S703. Therefore, when the filter coefficient is not changed, no substantial processing is performed. However, even if the setting of the filter coefficient is changed, the original process of passing only the component of one sound signal at that frequency is performed, so it is not always necessary to return to the original setting of the filter coefficient. Absent. Then, the sound output device 2 executes the processes after step S103 shown in the sound output process of the first embodiment.

  Since other configurations and processes are the same as those in the first embodiment, the first embodiment will be referred to and description thereof will be omitted.

  FIG. 24 is an explanatory diagram conceptually showing an example of processing of the passage mechanism 22 provided in the sound output device 2 according to Embodiment 5 of the present invention. FIG. 24 is a graph showing the magnitude of the amplitude with respect to the frequency f at a certain point of time for each channel related to the sound signal. FIG. 24A shows the first channel ch1 (f), and FIG. The second channel ch2 (f) and FIG. 24C show the nth channel chn (f), respectively. Since the correlation between the channels is high at the frequency f1, the passage mechanism 22 performs processing based on a preset filter coefficient. At the frequency f2, since the correlation is low and the amplitude of the sound signal related to the first channel ch1 (f) is large, the filter coefficient of the band pass filter unit 221 related to the first channel ch1 (f) is 1.0. The filter coefficients of the band pass filter units 221, 221 and so on relating to the channel are 0.0. Further, at the frequency f3, since the correlation is low and the amplitude of the sound signal related to the second channel ch2 (f) is large, the filter coefficient of the band pass filter unit 221 related to the second channel ch2 (f) is 1.0. The filter coefficients of the band-pass filter units 221, 221. Furthermore, at the frequency f4, since the correlation is low and the amplitude of the sound signal related to the n-th channel chn (f) is large, the filter coefficient of the band-pass filter unit 221 related to the n-th channel chn (f) is 1.0. The filter coefficients of the band-pass filter units 221, 221.

Embodiment 6 FIG.
The sixth embodiment is a form in which the filter coefficient setting method is changed in the fifth embodiment.

  In the following description, the fifth embodiment or the first embodiment is referred to for the same configuration as the first embodiment that is the basis of the fifth embodiment or the fifth embodiment, and the description thereof is omitted. Since the configuration example of the echo suppression system in the sixth embodiment is the same as that in the first embodiment shown in FIG. 4, the first embodiment is referred to and the description thereof is omitted.

  FIG. 25 is a functional block diagram showing a configuration example of the passage mechanism 22 of the sound output device 2 according to Embodiment 6 of the present invention. As in the fifth embodiment, the comparison unit 224 included in the passing mechanism 22 has an amplitude larger than a predetermined value by a predetermined number of frames or a predetermined time interval. , And if it is determined that the sound signal is present, the frequency, that is, the deflection frequency fd is passed to each of the bandpass filter units 221, 221,.

  When the deflection frequency fd is received, the band pass filter units 221, 221,... Change the filter coefficient at the received deflection frequency fd to 1.0. In this way, by setting the filter coefficients of all the band pass filter units 221, 221... To 1.0, all the sound signals X1 (f),. Since the amplitudes of all other sound signals are negligibly small relative to the amplitude of one sound signal, even if all the sound signals X1 (f),..., Xn (f) are passed through, Thus, the same effect as in the fifth embodiment can be obtained.

  When it is determined by comparison of the comparison unit 224 that there is no one sound signal having an amplitude greater than a predetermined value with respect to the amplitudes of all other sound signals, the sound signals X1 (f),. It is judged that the correlation between f) is strong. When it is determined that the correlation is strong, the comparison unit 224 does not derive the deflection frequency fd. Therefore, the band pass filter units 221, 221,... Execute the same processing as in the first embodiment. In addition, not only the deflection frequency fd is passed from the comparison unit 224 to the bandpass filter units 221, 221,..., But the filter coefficients of the bandpass filter units 221, 221,. You may make it pass to filter part 221,221, .... In this case, all the filter coefficients passed from the comparison unit 224 to the band pass filter units 221, 221... Are 1.0.

  Next, processing of each device provided in the echo suppression system according to Embodiment 6 of the present invention will be described. FIG. 26 is a flowchart showing an example of sound output processing of the sound output device 2 according to Embodiment 6 of the present invention. In the sixth embodiment, the processes of steps S101 to S102 of the sound output process according to the first embodiment are executed. Then, the passing mechanism 22 of the sound output device 2 uses the comparison unit 224 to compare the amplitudes of the sound signals X1 (f),..., Xn (f) for each frequency band, and each sound signal X1 (f). ,..., Xn (f) are weak, and it is determined whether or not there is one sound signal having an amplitude greater than a predetermined value with respect to the amplitudes of all other sound signals (S801).

  In step S801, if the correlation between the sound signals X1 (f),..., Xn (f) is weak and there is one sound signal having an amplitude greater than a predetermined value with respect to the amplitudes of all other sound signals. If it is determined (S801: YES), the pass mechanism 22 of the sound output device 2 passes the deflection frequency fd from the comparison unit 224 to each of the band pass filter units 221, 221,.

  The pass mechanism 22 of the sound output device 2 changes the setting of the filter coefficient at the deflection frequency fd of each of the band pass filter units 221, 221, ... to 1.0 (S802). Then, the sound output device 2 executes the processes after step S103 shown in the sound output process of the first embodiment.

  In step S801, the correlation between the sound signals X1 (f),..., Xn (f) is strong, and there is no one sound signal having an amplitude greater than a predetermined value with respect to the amplitudes of all other sound signals. (S801: NO), the pass mechanism 22 of the sound output device 2 cancels the change of the filter coefficient setting of each of the band pass filter units 221, 221, ... (S803). Canceling the change in step S803 is a process of canceling the changed setting and returning it to the original filter coefficient setting when the filter coefficient setting has been changed in step S802. Therefore, when the filter coefficient is not changed, no substantial processing is performed. Then, the sound output device 2 executes the processes after step S103 shown in the sound output process of the first embodiment.

  As a process of step S103 executed after the processes of steps S801 to S803, the pass mechanism 22 of the sound output device 2 uses a plurality of sounds converted into components on the frequency axis by the band pass filter units 221, 221. The components X1_f (f),..., Xn_f (f) in different frequency bands are passed through the signals X1 (f),. Note that at each frequency f, there is only one sound signal passing through, but all the sound signals X1 (fd),..., Xn (fd) pass through the component of the deflection frequency fd.

  Since other configurations and processes are the same as those in the first embodiment, the first embodiment will be referred to and description thereof will be omitted.

Embodiment 7 FIG.
The seventh embodiment is a form in which the filter coefficient of each bandpass filter unit is changed over time in the first embodiment.

  In the following description, components similar to those in the first embodiment are denoted by the same reference numerals as those in the first embodiment, and detailed description thereof is omitted. Since the configuration example of the echo suppression system in the seventh embodiment is the same as that in the first embodiment shown in FIG. 4, the first embodiment is referred to and the description thereof is omitted.

  FIG. 27 is a functional block diagram showing a configuration example of the passing mechanism 22 of the sound output device 2 according to Embodiment 7 of the present invention. The pass mechanism 22 according to the seventh embodiment includes a coefficient unit 225 that changes the filter coefficient of each bandpass filter unit 221, 221.

  The coefficient unit 225 has a clock circuit (not shown) that acquires time, and derives filter coefficients of the band-pass filter units 221, 221,... At a predetermined time interval. The coefficient unit 225 derives 1.0 as the filter coefficient of one band pass filter unit 221, and derives 0.0 as the filter coefficient of the other band pass filter units 221, 221,. Note that the band-pass filter unit 221 having a filter coefficient of 1.0 changes every time. The coefficient unit 225 derives the filter coefficient by various methods such as using a table in which the filter coefficient is recorded in advance in association with time and using a predetermined function that outputs the filter coefficient. Then, the derived filter coefficients of the band pass filter units 221, 221,... Are passed to the corresponding band pass filter units 221, 221,.

  In the band pass filter units 221, 221 and so on, the filter coefficients are changed to the accepted filter coefficients. The coefficient unit 225 derives filter coefficients that are changed every time at a predetermined time interval, so that the filter coefficients of the bandpass filter units 221, 221,... Are changed at a predetermined time interval, that is, with time. It will be. Further, the passing mechanism 22 changes the frequency band component for each sound signal to be passed with time. Instead of changing the filter coefficient based on the time indicated by the clock circuit, the filter coefficient may be changed over time by changing the filter coefficient every predetermined number of frames.

  Next, processing of each device provided in the echo suppression system according to Embodiment 7 of the present invention will be described. FIG. 28 is a flowchart showing an example of coefficient change processing of the sound output device 2 according to Embodiment 7 of the present invention. The passage mechanism 22 of the sound output device 2 uses the coefficient unit 225 to refer to the clock circuit and determine whether a predetermined time set in advance has elapsed (S901).

  If it is determined in step S901 that the predetermined time has elapsed (S901: YES), the pass mechanism 22 of the sound output device 2 derives the filter coefficients of the band pass filter units 221, 221. (S902), the derived filter coefficients are passed to the corresponding band-pass filter units 221, 221. If it is determined in step S901 that the predetermined time has not elapsed (S901: NO), the process of step S901 is repeated.

  The pass mechanism 22 of the sound output device 2 changes the setting of the filter coefficient of each band pass filter unit 221, 221... (S903). By the coefficient changing process, the passing mechanism 22 changes the frequency band component for each sound signal to be passed with time.

  Since other configurations and processes are the same as those in the first embodiment, the first embodiment will be referred to and description thereof will be omitted.

  FIG. 29 is an explanatory diagram illustrating an example of a change over time of the filter coefficients of the band pass filter units 221, 221... Of the pass mechanism 22 included in the sound output device 2 according to Embodiment 7 of the invention. FIGS. 29A, 29B, and 29C show the sound signals X1 (f), X2 (f), and Xn (f) of the first channel 1ch, the second channel 2ch, and the nth channel nch, respectively. The coefficient filters C1 (f), C2 (f), and Cn (f) are shown, and the horizontal axis is the frequency f and the vertical axis is the filter coefficient. 29 (d), (e), and (f) show the first channel 1ch, the second channel 2ch, and the second channel after a predetermined time has elapsed from the states shown in FIGS. 29 (a), (b), and (c), respectively. The coefficient filters C1 (f), C2 (f), and Cn (f) for the sound signals X1 (f), X2 (f), and Xn (f) of the n-th channel nch are shown. As is clear from comparison of FIGS. 29A, 29B and 29C with FIGS. 29D, 29E and 29F, the value of the coefficient filter of each channel, that is, the filter coefficient is a function of time. Has been changed. And the component of the frequency band for every sound signal to pass changes with time.

  The first to seventh embodiments are merely examples of an infinite embodiment of the present invention, and various hardware and software configurations can be set as appropriate. Various processes can be combined in addition to the basic processes. For example, the echo suppression system of the present invention can be applied to various systems related to voice and sound other than navigation systems and video conference systems, and further, echo suppression devices, sound output devices, sound input devices, and sound processing All of the devices or two or three devices appropriately selected may be configured as one device. Also, one device may be configured as a plurality of devices, such as dividing the sound output device into a sound signal generating device that generates a sound signal and a sound output device. Further, in the first to seventh embodiments, the form in which the comb filter is formed by setting the filter coefficient to 0.0 or 1.0 has been shown. It is possible to form filters having various characteristics such as a mold filter and a triangular filter. The first to seventh embodiments are not realized independently, but can be appropriately combined.

Claims (10)

A sound output device that outputs a sound based on a sound signal, a sound input device that converts an input sound into a sound signal, and a sound that is input from the sound input device based on a sound that is output from the sound output device In an echo suppression system comprising an echo suppression device for suppressing echo,
The sound output device is
For a plurality of sound signals, passing parts that pass components in different frequency bands, and
A plurality of sound output sections for outputting sounds based on the plurality of sound signals that have passed through the passage section,
An adder that generates a reference sound signal by adding a plurality of sound signals that have passed through the passage;
The echo suppressor is
An input unit for inputting a sound signal from the sound input device as an observation sound signal;
An echo suppression system, comprising: a correction unit that corrects the observation sound signal so as to suppress an echo included in the observation sound signal based on the observation sound signal and the reference sound signal. The passage part is
A first converter for converting a plurality of sound signals into components on the frequency axis,
Band pass filter units that pass components of different frequency bands for a plurality of sound signals converted into components on the frequency axis,
2. A second conversion unit that converts a plurality of sound signals converted into components on a frequency axis through which components of respective frequency bands are passed, into sound signals on the time axis, respectively. 2. The echo suppression system according to 1. The plurality of sound signals through which the passing sections pass components of different frequency bands are processed sound obtained by processing the first sound signal, the second sound signal, the first sound signal and / or the second sound signal by a predetermined processing method. Signal,
The band pass filter unit has a filter coefficient set for each sound signal,
Processing based on the first filter coefficient for the first sound signal, the second filter coefficient for the second sound signal, and the first filter coefficient and / or the second filter coefficient corresponding to the processing method for the processed sound signal The echo suppression system according to claim 2, wherein filter coefficients are respectively set. The sound output device is
The frequency band component corresponding to the preset removal band is not allowed to pass through any sound signal,
The correction unit is
A filter unit for correction for deriving a correction amount required for correcting the observation sound signal by filtering the reference sound signal with a filter coefficient set for each frequency;
A coefficient updating unit for calculating and updating the filter coefficient of the correction filter based on the corrected observation sound signal;
The echo according to any one of claims 1 to 3 , further comprising: an update availability determination unit that determines whether the coefficient update unit can update based on a component of a removal band of the observation sound signal. Suppression system. The sound output device further includes:
For each frequency band, the amplitudes of multiple sound signals are compared using a predetermined method, and the result of comparison between the amplitude of one sound signal with the maximum amplitude and the amplitudes of all other sound signals A comparison unit for determining whether or not the value indicating is equal to or greater than a predetermined value;
The said passage part is made to pass only said one sound signal with respect to the frequency band which the said comparison part determined to be more than predetermined value. The any one of Claim 1 thru | or 4 characterized by the above-mentioned. The echo suppression system described in 1. The sound output device further includes:
For each frequency band, the amplitudes of multiple sound signals are compared using a predetermined method, and the result of comparison between the amplitude of one sound signal with the maximum amplitude and the amplitudes of all other sound signals A comparison unit for determining whether or not the value indicating is equal to or greater than a predetermined value;
The said passage part is made to pass all the sound signals with respect to the frequency band which the said comparison part determined to be more than predetermined value. The Claim 1 thru | or 5 characterized by the above-mentioned. Echo suppression system. A sound output device that outputs sound based on a sound signal, a sound input device that receives sound, and an echo that suppresses echo based on the sound output from the sound output device from the sound input by the sound input device In an echo suppression method using a suppressor,
By the sound output device,
A procedure for passing components of different frequency bands for a plurality of sound signals,
A sound output procedure for outputting sounds based on a plurality of sound signals through which components of different frequency bands have passed,
With the sound input device,
A sound input procedure for converting the input sound into a sound signal;
By the echo suppression device,
An addition procedure for generating a reference sound signal by adding a plurality of sound signals that have passed through components of different frequency bands;
An input procedure in which a sound signal is input as an observation sound signal from the sound input device;
An echo suppression method, comprising: a correction procedure for correcting the observation sound signal so as to suppress an echo included in the observation sound signal based on the observation sound signal and the reference sound signal. In cooperation with a sound output device that outputs a sound based on a sound signal and a sound input device that generates a sound signal based on an input sound, the sound input device outputs from the input sound. In an echo suppression program for causing an echo suppressor to execute a procedure for suppressing echo based on the sound
In the echo suppressor,
A reference sound signal generated by adding a plurality of sound signals that have passed through components of different frequency bands output from the sound output device, and an observation sound signal that is a sound signal generated by the sound input device; And executing a correction procedure for correcting the observation sound signal so as to suppress the echo contained in the observation sound signal. In cooperation with a sound output device that outputs a sound based on a sound signal and a sound input device that generates a sound signal based on an input sound, the sound input device outputs from the input sound. In an echo suppressor that suppresses echo based on the generated sound,
An adder that generates a reference sound signal by adding a plurality of sound signals that have passed through components of different frequency bands output from the sound output device;
An input unit for inputting a sound signal from the sound input device as an observation sound signal;
An echo suppression apparatus comprising: a correction unit that corrects the observation sound signal so as to suppress an echo included in the observation sound signal based on the observation sound signal and the reference sound signal. A sound input device that includes a plurality of sound output units that output a sound based on a plurality of sound signals, generates a sound signal based on the input sound, and the sound output unit from the sound input to the sound input device In a sound output device that cooperates with an echo suppressor that suppresses echo based on sound output from
For a plurality of sound signals, passing parts that pass components in different frequency bands, and
An adder that generates a reference sound signal by adding a plurality of sound signals that have passed through the passage;
With
Wherein the plurality of sound output section, Ri said sound based on the plurality of sound signals passed through the passing section configured so as to respectively output Thea,
The addition unit, a sound output device, wherein the generated reference sound signal that Ru Thea configured so as to be output to the echo suppressor.

Images