JP2003124849A - Echo canceler and its method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the articulation of a transmit voice by reducing the influence of specific noise on a near-end side and preventing the echo removing capability of an echo canceler from decreasing in the case that there is the specific noise on the near-end side. SOLUTION: The echo canceler is equipped with a noise signal dictionary in which specific noise signals (winker sound, key input sound, etc.), as objects to be removed are previously gathered and recorded, a specific noise sensor which electrically transmits a noise production signal simultaneously with transmittance of a driving signal for a specific noise source, and an artificial noise generation part which generates artificial noise referring to the noise signal dictionary, and does not acoustically estimate the production of specific nose, but electrically or physically receive information showing that the noise is definitely produced to increase the removal precision of the specific noise.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an echo canceling technique used in a mobile terminal, a car telephone and the like, and more particularly to an echo canceller capable of removing a specific noise among ambient noises and a method thereof. 2. Description of the Related Art As a car phone system, a hands-free phone system is known in which a call is made using a microphone (speaker) and a speaker (receiver) that are arranged in advance in a specific position in the vehicle.

In this hands-free telephone system, in addition to the acoustic echo generated by the acoustic coupling between the microphone and the speaker, ambient noise such as running sound / horn sound / wiper sound / winker sound is generated between the transmitting side and the receiving side. It is mixed in between and deteriorates the call quality. In addition, the echo cancellation technology can be used to remove sneak from the speaker to the microphone even when the device (system) that outputs sound as well as communication is equipped with a voice input for operating the device using voice recognition. Used.

For example, in a car navigation system, "Where is your destination?""Hachioji? Enter yes if correct and no if different."
When voice guidance is given, a wraparound cancel technology is used to accept the user's answer at the same time as the guidance and correctly recognize the answer.

[0004]

2. Description of the Related Art FIGS. 10 and 11 show an example of the configuration of an echo canceller according to the prior art (Japanese Patent Laid-Open No. 2000-49668). In FIG. 10, 1 is an echo canceller unit used to eliminate acoustic echo and ambient noise, 2 is a speaker (receiver), 3 is a microphone (transmitter), 4 is ambient noise (running sound / running sound / A noise source such as a horn sound / wiper sound), and 11 is a specific noise detection unit that detects whether or not the specific noise to be removed is generated.

In FIG. 10, the input sound is output to the speaker 2 and is also input to the echo canceller unit 1 as a reference signal. On the other hand, the microphone 3 receives the voice of the near-end speaker, and ambient noise that includes the voice circling from the speaker 2 (hereinafter referred to as acoustic echo) and the noise from the near-end specific noise source 4 (hereinafter referred to as specific noise). Is also entered.

Among the near-end voice, acoustic echo, specific noise, and other noises, other noises are always input to the microphone 3. Here, the echo canceller unit 1 continues to monitor the input sound, and the information of the nature of the voice held by itself (for example,
Occurrence of the received voice is detected by using the spectrum and level fluctuation). Similarly, the microphone input sound is continuously monitored, and the occurrence of the near-end voice and the specific noise is detected by using the information of the characteristics of the voice and the specific noise held by the respective microphones.

First, acoustic echo removal will be described.
Acoustic echo removal is performed when the echo canceller unit 1 detects an input sound. In this case, the echo canceller unit 1 refers to the input sound, creates a pseudo echo using information on the transmission path of the acoustic echo (for example, tap coefficient), and subtracts the acoustic echo from the microphone input signal. Remove.

Next, learning of the transmission path of the acoustic echo will be described. Learning is performed when the input sound is detected by the echo canceller unit 1 and no voice or specific noise is detected in the microphone input sound, that is, when the microphone input signal includes only the acoustic echo component. In this case, the acoustic echo is also removed by the above method. The transmission signal after the removal is an unerased component of the acoustic echo, and the information is used to update the information on the transmission path of the acoustic echo.

Next, the specific noise removal will be described. The specific noise removal is performed when the specific noise detector 11 detects the specific noise. In this case, the echo canceller unit 1 refers to the specific noise signal held by the echo canceller unit 1, creates pseudo noise using information on the transmission path of the specific noise, and subtracts it from the microphone input sound to remove the specific noise. To do. Next, learning of the transmission path of specific noise will be described. The learning is performed when the specific noise detection unit 11 detects the specific noise and does not detect the voice in the microphone input sound, that is, when the microphone input sound includes only the specific noise component. In this case, the specific noise is also removed by the above method. The transmission signal after the removal is the unerased component of the specific noise, and the information is used to update the information of the transmission path of the specific noise.

FIG. 11 is a block diagram (part 2) of an echo canceller according to the prior art. Only differences from the configuration diagram (1) of the echo canceller according to the related art of FIG. 10 will be described below. In FIG. 11, reference numeral 12 is a microphone dedicated to specific noise detection. This specific noise detection microphone 12
Is the most likely to receive a specific noise, but it is difficult to receive a near-end voice or an acoustic echo.
It is arranged independently of.

The specific noise detecting section 11 continues to monitor the noise signal received from the noise detecting microphone 12, and detects the occurrence of the specific noise by using the information of the property of the specific noise held by itself. That is, when the specific noise detection unit 11 detects the specific noise, the specific noise detection unit 11 transfers the estimation information on the presence or absence of noise noise generation to the echo canceller unit 1. All other configurations are shown in FIG.
Is the same as.

As described above, the echo canceller unit 1 eliminates acoustic echoes and suppresses howling to realize bidirectional simultaneous communication.

[0013]

The conventional echo canceller device has the following problems. First,
The near-end ambient noise input from the microphone was heard by the far-end speaker, which was very annoying to the far-end speaker.
Secondly, when there is ambient noise on the near end side, the echo canceling capability of the echo canceller unit decreases, and the acoustic echo returns to the far end side, resulting in a decrease in call quality.

Thirdly, when an excessive input suddenly occurs during a call, the echo canceller cannot completely remove the acoustic echo, and the residual acoustic echo returns to the far end side, which deteriorates the call quality. . Also, when the acoustic echo cannot be completely removed, a loop is formed in the acoustic path at the near end and the far end of the echo, which causes howling. In the configuration of FIG. 10, the microphone input sound on which the near-end speaker's voice (near-end voice), acoustic echo, specific noise, and other noise (acoustic echo, noise other than specific noise) are superimposed is acoustically analyzed to determine the specific noise. The specific noise is detected by estimating whether or not is generated. Therefore, there is a possibility that the specific noise is erroneously detected, and if the erroneous detection is performed, the noise cannot be removed and the acoustic echo cannot be completely removed, and the residual acoustic echo returns to the far end side, resulting in deterioration of the call quality. .

In the configuration of FIG. 11, the detection processing is performed by estimating whether or not the specific noise is generated by acoustically analyzing the specific noise signal input from the noise detecting microphone. However, it is difficult to input only specific noise with the noise detection microphone, and in reality, the voice of the near-end speaker (near-end voice), acoustic echo, and other noises are likely to be superimposed. Therefore, as in the case of FIG. 10, the call quality is deteriorated due to the erroneous detection of the specific noise.

The present invention has been made in order to solve the above problems, and can reduce the influence of the specific noise on the near end side, and at the same time, can eliminate the echo by the echo canceller when the specific noise on the near end side is present. The purpose of this is to prevent the deterioration of speech and improve the clarity of transmitted voice.

[0017]

A first invention for solving the above-mentioned problems is an echo canceller for a system in which an acoustic echo generated by acoustic coupling between a microphone and a speaker and ambient noise are mixed. Equipped with a specific noise sensor that directly records a noise generation signal to the echo canceller unit in which the signal of the specific noise source to be removed is recorded in advance and stored in the echo canceller unit. ing.

According to the first aspect of the present invention, the noise generation signal can be obtained at the same time that the noise is generated, so that the presence or absence of the noise can be accurately obtained. As a result, the performance of removing the specific noise can be improved, and an echo canceller that prevents a decrease in the echo cancellation performance can be provided. A second invention is an echo canceller of a system in which an acoustic echo generated by acoustic coupling between a microphone and a speaker and ambient noise are mixed, and a signal of a specific noise source to be removed is previously recorded and stored. A noise signal dictionary, a specific noise sensor that directly transmits a noise generation signal simultaneously with the transmission of a drive signal that drives the specific noise source, and a specific noise of the noise signal dictionary based on a signal from the specific noise sensor A pseudo noise generating unit for generating a pseudo noise signal by referring to a signal, a pseudo echo generating unit for generating a pseudo echo signal by referring to a received signal, and a subtractor for subtracting the pseudo noise signal from the microphone input signal , A subtractor that subtracts the pseudo echo signal from the microphone input signal.

According to the second aspect of the present invention, the influence of the specific noise on the near end side can be reduced, and the echo canceling ability of the echo canceller can be prevented from being lowered when there is the specific noise on the near end side. It is possible to provide an echo canceller with improved clarity of a transmitted voice signal. A third invention is an echo canceller of a system in which an acoustic echo generated by acoustic coupling between a microphone and a speaker and ambient noise are mixed, and a signal of a specific noise source to be removed is previously recorded and stored. A noise signal dictionary, a specific noise sensor that directly transmits a noise generation signal simultaneously with the transmission of a drive signal that drives the specific noise source, and a specific noise of the noise signal dictionary based on a signal from the specific noise sensor A pseudo echo noise generating unit that refers to a signal to generate a pseudo noise signal, and refers to a received signal to generate a pseudo echo signal; and a subtractor that subtracts the pseudo echo noise signal from the microphone input signal. Is equipped with.

According to the third aspect of the present invention, the pseudo echo noise generator is configured so as to share the pseudo echo generating section for generating the pseudo echo signal and the pseudo noise generating section for generating the pseudo noise signal. Therefore, a more economical echo canceller can be provided. A fourth invention is an echo canceller of a system in which acoustic echo generated by acoustic coupling between a microphone and a speaker and specific ambient noise are mixed, and a signal of a specific noise source to be removed is recorded in advance. , A noise signal dictionary to be stored, a specific noise sensor for directly transmitting a noise generation signal simultaneously with the transmission of a drive signal for driving the specific noise source, and a pseudo echo generation for generating a pseudo echo signal by referring to a received signal A subtracter that reads a noise signal from the noise signal dictionary based on a signal from the specific noise sensor and subtracts the noise signal from the microphone input signal, and a subtractor that subtracts the pseudo echo signal from the microphone input signal. ing.

According to the fourth aspect of the invention, when the noise to be removed is a reproducible sound, for example, a mechanically generated sound, or the sound source signal and the sound source to the microphone are transmitted. When the path is unlikely to change, it is not necessary to calculate a pseudo noise signal for noise removal, and an echo canceller that can reduce the calculation cost can be provided.

A fifth aspect of the present invention is an echo canceling method for a system in which an acoustic echo generated by acoustic coupling between a microphone and a speaker and ambient noise are mixed, and a signal of a specific noise source to be removed is previously set. The step of recording and storing, the step of detecting the presence or absence of the received voice signal, the step of detecting the presence or absence of the voice of the near-end speaker, and the step of directly transmitting the noise generation signal simultaneously with the transmission of the drive signal for driving the specific noise source. A step of transmitting, a step of creating a pseudo noise signal by referring to the specific noise signal in the noise signal dictionary based on a signal from the specific noise sensor, and a step of creating a pseudo echo signal by referring to a received signal , Subtracting the pseudo noise signal and the pseudo echo signal from the microphone input signal.

According to the fifth aspect of the present invention, the influence of the ambient noise on the near end side can be reduced, and at the same time, it is possible to prevent the reduction of the echo removing ability for canceling the echo when there is ambient noise on the near end side. It is possible to provide an echo canceling method that improves the clarity of the transmitted voice signal.

[0024]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings. In addition, in order to facilitate understanding through all the drawings, the same portions are denoted by the same reference numerals. FIG. 1 is a configuration diagram of an echo canceller according to a first embodiment of the present invention, which is a drive signal for driving a specific noise source that causes noise generation (for example, a noise drive current or a signal for detecting noise generation). ) And an echo canceller that suppresses specific noise on the near end side by means of directly transmitting a noise generation signal (transmitting as an electric signal) at the same time.

In FIG. 1, 1 is an echo canceller unit used to eliminate acoustic echo and ambient noise, 2 is a speaker (receiver), 3 is a microphone (transmitter), and 4 is presumed ambient noise ( (Running sound / horn sound / wiper sound etc.)
Noise source, 5 is a specific noise sensor that detects whether or not the specific noise to be removed is generated by an electric signal, 51 is a drive source of the specific noise, and 52 is two switches that operate in conjunction with each other.
3 has shown the power supply.

In FIG. 1, the specific noise sensor 5 is composed of a specific noise driving source 51, a power source 53, and two switches 52, and the two switches 52 are configured to operate in conjunction with each other. Here, when the switch 52 is turned on, a drive signal is sent from the drive source 51 to the noise source 4, and noise is generated. At the same time, a noise generation signal is sent from the power supply 53 to the echo canceller unit 1. With this, it is possible to obtain the noise generation signal at the same time that the noise is generated, and it is possible to accurately obtain the presence or absence of the noise generation.

However, the drive source 51, the switch 52, and the noise source 4 are model diagrams, and do not always exist. Therefore, the present invention includes a mechanism in which the specific noise sensor physically detects the cause of the specific noise and transmits the noise generation signal. FIG. 2 shows the second aspect of the present invention.
FIG. 3 is a configuration diagram of an echo canceller in one embodiment, in which a specific noise sensor receives an electric signal from a power source or the like driving a noise source that causes noise,
The present invention shows an echo canceller in which the presence or absence of specific noise is known, resources for detecting the specific noise are not required, and there is no side effect (cancellation failure) due to detection failure.

Only differences from the configuration of the echo canceller according to the first embodiment of the present invention shown in FIG. 1 will be described below. In FIG. 2, reference numeral 101 denotes a pseudo echo generation unit that creates a pseudo echo by referring to a received signal, and 102 creates a pseudo noise signal by referring to a specific noise signal in a noise signal dictionary based on a signal from the specific noise detection unit. Pseudo noise generators 103, 103 ′ are subtractors for subtracting the pseudo echo signal / pseudo noise signal from the microphone input signal, 20
Reference numeral 1 is a voice detection unit that detects the presence or absence of a voice signal from the speaker output sound and the microphone input sound, and 202 is a control that issues a control signal for canceling acoustic echo and noise based on the detection results from the voice detection unit and the specific noise sensor. Reference numeral 203 denotes a noise signal dictionary in which noise signals are previously recorded and recorded.

In FIG. 2, the noise source 4 and the specific noise sensor 5 are electrically connected directly to each other, and a signal that causes noise generation (for example, a driving current of noise) or a signal that detects noise generation is transferred. It is supposed to be done. Further, the noise signal dictionary 203 preliminarily records and stores noise to be removed. In FIG. 2, the reception signal output from the speaker 2 is x, the signal of the noise source 4 is n, the acoustic echo signal obtained by transmitting from the speaker 2 to the microphone 3 is x ′, and the acoustic echo signal is transmitted from the noise source 4 to the microphone 3. The resulting signal is n ',
When the voice of the near-end speaker is s, the microphone input signal y ”
Becomes y ″ = s + x ′ + n ′.

In addition, a noise signal in which noise is recorded in advance is given as n _p
Then, first, the pseudo noise generating unit 102 refers to n _p to estimate the transfer characteristic from the noise source to the microphone,
Find n _p '. This was subtracted from y ", 'a. Here, n represents the ideal' y '= s + x' + n'-n p ' so the, y' ≒ n _p becomes ≒ s + x '.

Next, the pseudo echo generating unit 101 refers to x, estimates the transfer characteristic from the speaker 2 to the microphone 3, and obtains x _p '. Subtracting this from y ′ yields y = s + x′−x _p ′. Here, ideally, x ′ = x _p ′, so that y ′ = s.

By the above processing, the voice of the near-end speaker can be obtained as the transmission signal. Here, if the method of estimating the generation of noise to be removed from the sound waveform is involved in the calculation cost for the estimation, and if the estimation fails, not only the noise removal performance but also the echo removal performance is degraded. .
However, in the present invention, the specific noise sensor 5 can definitely detect the noise to be removed, which is very effective.

FIG. 3 is a block diagram of an echo canceller according to a third embodiment of the present invention, in which a pseudo echo generating section for generating a pseudo echo signal and a pseudo noise generating section for generating a pseudo noise signal are shared. 2 shows an echo canceller including the pseudo echo noise generator. Only the differences from the configuration diagram of the echo canceller in the first embodiment of the present invention in FIG. 1 and the configuration diagram of the echo canceller in the second embodiment of the present invention in FIG. 2 will be described below.

In FIG. 3, reference numeral 104 is a pseudo echo noise generator. This pseudo echo noise generation unit 104
Generates both a pseudo echo and a pseudo noise by referring to the received voice signal and the noise signal dictionary 203 in which noise to be removed is recorded and stored in advance. When the specific noise is generated from the noise source 4, a noise signal is transmitted from the noise signal dictionary 203 to obtain pseudo noise, and when there is no specific noise, the noise signal dictionary 203 is generated.
Outputs silence or does not emit a signal.

In the present configuration according to FIG. 3, the pseudo echo / noise generating section 104 is configured to share the pseudo echo generating section for generating the pseudo echo signal and the pseudo noise generating section for generating the pseudo noise signal. Therefore, a more economical echo canceller can be provided. FIG. 4 is a configuration diagram of an echo canceller according to a fourth embodiment of the present invention, and shows an echo canceller in which the calculation cost is reduced by not providing a pseudo noise generating unit that calculates pseudo noise.

The configuration of the echo canceller according to the first embodiment of the present invention shown in FIG. 1, the configuration of the echo canceller according to the second embodiment of the present invention shown in FIG. 2, and the third embodiment of the present invention shown in FIG. Only differences from the configuration diagram of the echo canceller according to the embodiment will be described. In FIG. 4, when the noise to be removed is generated from the noise source 4, the specific noise sensor 5 senses a signal (for example, a driving current of noise) that causes the noise or the noise generation. The signal is received, the presence or absence of noise is identified, and the noise detection signal is transferred to the control unit 202. The control unit 202 sends a control signal to the noise signal dictionary 203 based on the reception of the information on the presence / absence of occurrence of the specific noise. The noise signal dictionary 203 is
Noise removal is performed by referring to a specific noise signal to be removed, which is recorded and stored in advance, and subtracting the specific noise signal from the input signal of the microphone 3 by the subtractor 103 ′. You can

In the present configuration shown in FIG. 4, when the noise to be removed is a reproducible sound, for example, when it is a mechanically generated sound, the noise source signal and the noise source 4 to the microphone 3 When the transmission path is difficult to change, it is not necessary to calculate the pseudo noise signal for noise removal, and the calculation cost can be reduced. That is, it is not necessary for the pseudo noise generating unit to learn the transfer characteristic from the noise source to the microphone, so that the pseudo noise generating unit can be omitted.

FIG. 5 is a configuration diagram in the case where the turn signal driving sound is applied as the noise source in FIG. 2, and shows an echo canceller for suppressing the "clicking" noise generated by the turn signal driving in the car telephone system. . In FIG. 5, 6 is a winker that interlocks with a turn signal indicator of an automobile, 7 is a winker lever, 8 is a winker 6, and the winker lever 7 is electrically directly connected to a winker drive sensor that detects a driving current of the winker.
Reference numerals 05 and 105 ′ are adaptive filters normally formed by a transversal filter, and switches 1 to 4
Is a switch for turning on / off or switching a connection route for controlling learning and subtraction of echo and noise, and 202 is an adaptive filter 105 and 105 ′ according to detection states of the guide voice, voice input, and specific noise. A control unit for controlling the switch 1, the switch 2, the switch 3, and the switch 4 for switching the connection of the switch, 204 is a turn signal driving sound dictionary in which the turn signal noise to be removed is recorded and stored in advance.

Normally, when interacting with the car navigation system while operating the vehicle (by operating with a voice), the turn signal drive notification sound (sound "click") generated when driving the turn signal is recognized by the car navigation. At the time of execution, it was a cause of a decrease in recognition rate. Here, the turn signal driving sound is an unsteady sound,
It is difficult to remove with a conventional noise canceller, which is one of the causes for lowering the echo removal performance in the conventional echo canceller unit.

In the embodiment of the present invention shown in FIG. 5, a drive signal is received from the winker drive sensor 8 and noise removal processing is performed during driving. In this example, the car navigation system is assumed to be standard equipment (built-in type) of the vehicle, and the vehicle side is provided with an interface for notifying the car navigation system of winker driving information. The operation flow of FIG. 6 will be described below with reference to FIG.

S1. Receive the guidance text by voice from the car navigation system. S2. The voice detection unit 201 and the winker drive sensor 8 are
The state (presence or absence of signal) of each of the voice guidance sentence, the voice input of the interlocutor, and the noise signal is detected. S3. In the control unit 202, each state of the guidance sentence, the voice input, and the noise signal corresponds to any one of the state 1, the state 3, the state 4, and the state 7 shown in the operation switching specification table (Table 1). Or not.

[0042]

[Table 1] S4. As shown in the operation switching specification table (Table 1), the pseudo noise is calculated by turning on the switch 4. S5. The pseudo noise is subtracted from the voice input signal in which the acoustic echo, the voice input to the microphone 3, and the noise are mixed by the subtractor 103 via the switch 4 from the adaptive filter 105 ′.

S6. In the control unit 202, each state of the guidance sentence, voice input, and noise signal corresponds to any one of state 1, state 2, state 3, and state 5 shown in the operation switching specification table (Table 1). Or not. S7. As shown in the operation switching specification table (Table 1), the switch 2 is turned on to calculate the pseudo echo.

S8. The pseudo echo is the adaptive filter 105.
Via switch 2 through subtractor 103 to microphone 3
The pseudo echo is subtracted from the audio input signal with the acoustic echo, audio input, and noise to the. S9. The control unit 202 determines whether or not each state of the guidance sentence, voice input, and noise signal corresponds to state 7 shown in the operation switching specification table (Table 1).

S10. By ON-controlling the switch 3, a signal obtained by subtracting pseudo noise from the microphone input sound is input and the tap coefficient for pseudo noise of the adaptive filter 105 ′ is updated. S11. The control unit 202 determines whether or not each state of the guidance sentence, voice input, and noise signal corresponds to state 5 shown in the operation switching specification table (Table 1).

S12. By ON-controlling the switch 1, a signal obtained by subtracting the pseudo echo from the microphone input sound is input, and the tap coefficient for the pseudo echo of the adaptive filter 105 is updated. S13. The input voice is transmitted to a voice recognition unit (not shown) of the car navigation system.

As described above, FIG. 5 shows the case where it is applied to FIG. 2 which is the second embodiment of the present invention. Of course, when it is applied to FIG. 3 which is the third embodiment of the present invention, It goes without saying that the present invention can be realized by applying it to FIG. 4 which is the fourth embodiment of the present invention. FIG. 7 is a configuration diagram in the case where the key input sound of the mobile phone terminal is applied as the noise source in FIG.
Key input sound generated by key input during a hands-free call on a mobile phone (sound "click")
2 shows an echo canceller that removes.

In FIG. 7, reference numeral 9 is a mobile phone terminal, and 10 is a key input sensor which is electrically directly connected to the key input section of the mobile phone terminal 9 and detects a drive current thereof.
Indicates a key input sound dictionary in which each key input sound is recorded in advance and a sound waveform is recorded. Depending on the telephone service content,
There may be cases where key input is required during a hands-free call on a mobile phone, but this key input sound (sound "clicking") is transmitted as noise to the other party during a conversation, for example, and the communication quality Was the cause of the decrease.

Here, the key input sound is an unsteady sound, which is difficult to remove by the conventional noise canceller, which is one of the causes for lowering the echo removing performance in the conventional echo canceller section. In the embodiment of the present invention shown in FIG. 7, a key input drive signal is received from the key input unit of the mobile phone terminal 9, and noise removal processing is performed at the time of key input. In this example, it is assumed that an echo canceller unit is built in the mobile phone terminal.

FIG. 8 is an explanatory diagram of the key input sound dictionary shown in FIG. 7, in which the key “1” input sound waveform, the key “2” input sound waveform, corresponding to the address (identification signal) of the key input dictionary.
... Key "0" input sound waveform, key "*" input sound waveform,
Key “#” input sound waveform is stored. The operation flow of FIG. 9 will be described below with reference to FIG. 7.

S14. Receive the received voice signal. S15. The voice detection unit 201 and the key input sensor 10 detect the received voice signal, the voice of the near-end interlocutor, and the presence / absence of key input. S16. The control unit 202 determines whether the received voice signal, the near-end voice input, and the noise signal are in any one of state 1, state 3, state 4, and state 7 shown in the operation switching specification table (Table 2). Is determined.

[0052]

[Table 2] S17. As shown in the operation switching specification table (Table 2),
When the switch 7 is turned on, the ID of the input key is acquired. S18. Pseudo noise for ID is the key input sound dictionary 2
From 05, via the switch 7, the subtractor 103 ′ subtracts the pseudo noise from the near-end voice input signal in which the acoustic echo to the microphone 3, the near-end voice, and the noise are mixed.

S19. The control unit 202 controls the received voice signal,
It is determined whether or not each state of the near-end voice input and the noise signal corresponds to any one of state 1, state 2, state 3 and state 5 shown in the operation switching specification table (Table 2). . S20. As shown in the operation switching specification table (Table 2),
When the switch 6 is turned on, the pseudo echo is calculated.

S21. The pseudo echo is the adaptive filter 10
The pseudo echo is subtracted from the voice input signal in which the acoustic echo to the microphone 3, the near-end voice, and the noise are mixed by the subtractor 103 from 5 via the switch 6. S22. The control unit 202 determines whether or not each state of the received voice signal, the near-end voice input, and the noise signal corresponds to state 5 shown in the operation switching specification table (Table 2).

S23. Only the received voice is input, the echo is learned, and the pseudo echo tap coefficient of the adaptive filter 105 is updated. S24. A near-end voice signal from which acoustic echo and specific noise (key input sound) have been removed is transmitted to the other party. As described above, FIG. 7 shows the case where it is applied to FIG. 4 which is the fourth embodiment of the present invention, but naturally, when applied to FIG. 2 which is the second embodiment of the present invention, It is needless to say that the present invention can be realized by applying it to FIG.

(Supplementary Note 1) An echo canceller of a system in which an acoustic echo generated by acoustic coupling between a microphone and a speaker and ambient noise are mixed, in which a signal of a specific noise source to be removed is recorded in advance, An echo canceller, comprising: a specific noise sensor for directly transmitting a noise generation signal to the stored echo canceller section at the same time as transmitting a drive signal for driving the specific noise source. (Supplementary Note 2) An echo canceller for a system in which acoustic echo generated by acoustic coupling between a microphone and a speaker and ambient noise are mixed, in which a signal of a specific noise source to be removed is recorded and stored in advance. Noise signal dictionary to put,
A specific noise sensor that directly transmits a noise generation signal simultaneously with the transmission of a drive signal that drives the specific noise source, and based on a signal from the specific noise sensor, refers to the specific noise signal in the noise signal dictionary, and a pseudo noise signal For generating a pseudo echo signal by referring to a received signal, a subtractor for subtracting the pseudo noise signal from the microphone input signal, and a pseudo echo signal for the microphone. An echo canceller comprising: a subtractor that subtracts from an input signal. (Supplementary Note 3) An echo canceller for a system in which acoustic echo generated by acoustic coupling between a microphone and a speaker and ambient noise are mixed, in which a signal of a specific noise source to be removed is recorded in advance and stored. Noise signal dictionary to put,
A specific noise sensor that directly transmits a noise generation signal simultaneously with the transmission of a drive signal that drives the specific noise source, and based on a signal from the specific noise sensor, refers to the specific noise signal in the noise signal dictionary, and a pseudo noise signal Together with
An echo canceller, comprising: a pseudo echo noise generating unit that creates a pseudo echo signal by referring to a received signal; and a subtractor that subtracts the pseudo echo noise signal from the microphone input signal. (Supplementary note 4) An echo canceller of a system in which acoustic echo generated by acoustic coupling between a microphone and a speaker and specific ambient noise are mixed, and a signal of a specific noise source to be removed is previously recorded and stored. A noise signal dictionary, a specific noise sensor that directly transmits a noise generation signal simultaneously with the transmission of a drive signal that drives the specific noise source, and a pseudo echo generation unit that creates a pseudo echo signal by referring to a received signal A subtractor for reading a noise signal from the noise signal dictionary based on a signal from the specific noise sensor and subtracting it from the microphone input signal; and a subtractor for subtracting the pseudo echo signal from the microphone input signal. Characteristic echo canceller. (Supplementary note 5) A method of canceling an echo of a system in which acoustic echo generated by acoustic coupling between a microphone and a speaker and ambient noise are mixed, wherein a signal of a specific noise source to be removed is recorded and stored in advance. A step of detecting the presence or absence of a received voice signal, a step of detecting the presence or absence of a voice of a near-end speaker, a step of directly transmitting a noise generation signal simultaneously with the transmission of the drive signal for driving the specific noise source, Based on the signal from the specific noise sensor, referring to the specific noise signal in the noise signal dictionary, creating a pseudo noise signal, creating a pseudo echo signal by referring to the received signal, the pseudo noise signal And a step of subtracting the pseudo echo signal from the microphone input signal. (Supplementary Note 6) The specific noise source is a turn signal driving sound, and the noise signal dictionary is a turn signal driving sound dictionary.
The echo canceller according to any one of appendices 1 to 3, wherein the specific noise detector is a winker drive detector.

(Supplementary Note 7) The specific noise source is a key input sound of the mobile phone terminal, the noise signal dictionary is a key input sound dictionary, and the specific noise detection unit is a key input detection unit. The echo canceller according to any one of appendices 1 to 3. (Supplementary Note 8) The specific noise source is a turn signal driving sound, and the noise signal dictionary is a turn signal driving sound dictionary.
5. The echo canceling method according to appendix 4, wherein the specific noise detection unit is a winker drive detection unit.

(Supplementary Note 9) The specific noise source is a key input sound of the mobile phone terminal, the noise signal dictionary is a key input sound dictionary, and the specific noise detection unit is a key input detection unit. The echo canceling method according to attachment 4. (Supplementary note 10) The echo canceller according to any one of Supplementary notes 2 to 4, Supplementary note 6 and Supplementary note 7, wherein the noise signal dictionary includes a plurality of specific noise signals.

(Supplementary Note 11) Supplementary notes 5 and 8 characterized in that the noise signal dictionary includes a plurality of specific noise signals.
And the echo canceling method according to any one of Appendix 9.

[0060]

As described above, according to the present invention, the influence of the specific noise on the near end side can be reduced and the reduction of the echo canceling ability of the echo canceller in the presence of the specific noise on the near end side can be prevented. An echo canceller and an echo cancellation method with improved signal clarity can be provided.

[Brief description of drawings]

FIG. 1 is a configuration diagram of an echo canceller according to a first embodiment of the present invention.

FIG. 2 is a configuration diagram of an echo canceller according to a second embodiment of the present invention.

FIG. 3 is a configuration diagram of an echo canceller according to a third embodiment of the present invention.

FIG. 4 is a configuration diagram of an echo canceller according to a fourth embodiment of the present invention.

FIG. 5 is a configuration diagram when a turn signal driving sound is applied as a noise source in FIG.

FIG. 6 is a flowchart of FIG.

FIG. 7 is a configuration diagram when a key input sound of a mobile phone terminal is applied as a noise source in FIG.

8 is an explanatory diagram of a key input sound dictionary shown in FIG. 7. FIG.

9 is a flowchart of FIG. 7.

FIG. 10 is a configuration diagram (1) of an echo canceller according to a conventional technique.

FIG. 11 is a configuration diagram (2) of an echo canceller according to a conventional technique.

[Explanation of symbols]

1 Echo canceller section 2 speakers 3 microphone 4 noise sources 5 Specific noise sensor 6 turn signals 7 blinker lever 8 blinker drive sensor 9 Mobile phone terminals 10 key input sensor 11 Specific noise detector 12 Microphone for specific noise detection 51 drive source 52 Interlocking switch 53 power supply 101 Pseudo echo generator 102 Pseudo noise generator 103 Subtractor 104 Pseudo echo noise generator 105 Adaptive filter 201 voice detector 202 control unit 203 Noise signal dictionary 204 Turn signal driving sound dictionary 205 key input sound dictionary

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Hitoshi Sasaki             4-1, Kamiodanaka, Nakahara-ku, Kawasaki-shi, Kanagawa             No. 1 within Fujitsu Limited F term (reference) 5D020 BB07 CC00                 5K027 AA11 BB03 DD10 DD18                 5K046 AA05 BB01 HH11 HH71 HH78                       HH79

Claims (5)

[Claims] 1. An echo canceller for a system in which acoustic echo generated by acoustic coupling between a microphone and a speaker and ambient noise are mixed, wherein a signal of a specific noise source to be removed is recorded and stored in advance. An echo canceller, comprising: a specific noise sensor for directly transmitting a noise generation signal simultaneously with transmission of a drive signal for driving the specific noise source, to the echo canceller section. 2. An echo canceller for a system in which an acoustic echo generated by acoustic coupling between a microphone and a speaker and ambient noise are mixed, wherein a signal of a specific noise source to be removed is recorded and stored in advance. A noise signal dictionary, a specific noise sensor that directly transmits a noise generation signal simultaneously with the transmission of the drive signal that drives the specific noise source, and a specific noise signal of the noise signal dictionary based on the signal from the specific noise sensor , A pseudo noise generator that creates a pseudo noise signal, a pseudo echo generator that creates a pseudo echo signal by referring to a received signal, a subtractor that subtracts the pseudo noise signal from the microphone input signal, An echo canceller, comprising: a subtractor that subtracts the pseudo echo signal from the microphone input signal. 3. An echo canceller for a system in which acoustic echo generated by acoustic coupling between a microphone and a speaker and ambient noise are mixed, wherein a signal of a specific noise source to be removed is recorded and stored in advance. A noise signal dictionary, a specific noise sensor that directly transmits a noise generation signal simultaneously with the transmission of the drive signal that drives the specific noise source, and a specific noise signal of the noise signal dictionary based on the signal from the specific noise sensor And a pseudo echo noise generator that creates a pseudo echo signal by referring to the received signal and a subtractor that subtracts the pseudo echo noise signal from the microphone input signal. An echo canceller characterized by being provided. 4. An echo canceller for a system in which an acoustic echo generated by acoustic coupling between a microphone and a speaker and specific ambient noise are mixed, wherein a signal of a specific noise source to be removed is recorded in advance, A noise signal dictionary to be stored, a specific noise sensor that directly transmits a noise generation signal simultaneously with the transmission of the drive signal that drives the specific noise source, and a pseudo echo generation unit that creates a pseudo echo signal by referring to the received signal And a subtractor that reads a noise signal from the noise signal dictionary based on a signal from the specific noise sensor and subtracts the noise signal from the microphone input signal, and a subtracter that subtracts the pseudo echo signal from the microphone input signal. Echo canceller featuring. 5. A method of canceling an echo of a system in which an acoustic echo generated by acoustic coupling between a microphone and a speaker and ambient noise are mixed, wherein a signal of a specific noise source to be removed is recorded and stored in advance. A step of detecting the presence / absence of a received voice signal, a step of detecting the presence / absence of a voice of a near-end speaker, and a step of directly transmitting a noise generation signal simultaneously with the transmission of a drive signal for driving the specific noise source, A step of creating a pseudo noise signal by referring to a specific noise signal in the noise signal dictionary based on a signal from the specific noise sensor; a step of creating a pseudo echo signal by referring to a received signal; A signal and the pseudo echo signal from the microphone input signal.