JP2003110690A - Hands-free speech system, blocking preventing method, and blocking preventing program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To easily and readily realize the stability of a hands-free function of a voice communication device and the improvement of the performance of the device. SOLUTION: A hands-free speech system includes; at least echo cancellers that update a filter coefficient according to echo and generate a dummy echo so as to eliminate the echo, a voice switch 1 that detects a power signal of a transmission reception path and inserts an attenuation to either of the transmission reception paths on the basis of the power signal so as to select the transmission or the reception, a howling detector 8 for detecting howling on the basis of the power signal, and a blocking detector 9 for detecting blocking on the basis of the number of times of switching of the voice switch 1 within a specified time.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a voice communication system having a hands-free function, and more particularly to a blocking prevention method and a blocking prevention program.

[0002]

2. Description of the Related Art In recent years, a hands-free function of voice communication equipment has become widespread. In a voice communication device having a hands-free function, an echo caused by a sneak from a speaker to a microphone becomes a problem. That is, an acoustic echo is generated by the convolution of the actual voice and the impulse response, so-called acoustic coupling. Further, a line echo that returns as an echo is generated on the line side due to sidetone coupling on the transmission path.

Under these circumstances, as shown in FIG. 7, the conventional hands-free communication system has a sound side echo canceller 2 and a line side echo canceller 3 on the sound side and the line side, respectively. Further, in the conventional hands-free call system, the voice switch 1 is combined as an auxiliary system. The acoustic echo canceller 2 generates a pseudo echo signal by using an adaptive signal processing technique to remove the acoustic echo. The line echo canceller 3 also removes the line echo on the same principle. This adaptive signal processing is
It is mainly performed by the adaptive filters 13a and 13b. The filter coefficient updating means 14a and 14b update (learn) the filter coefficients of the adaptive filters 13a and 13b according to the acoustic echo and the line echo, thereby making the pseudo echo signal close to the acoustic echo and the line echo.

The voice switch 1 functions as a switch for switching a two-wire line to a single transmission / reception single (half duplex) mode. Actually, the switch is performed by inserting the attenuation amount into either the transmission path or the reception path. That is, according to the power signal component of the transmission / reception, the attenuation amount control means 10 causes the reception path side attenuation means 11a and the transmission side attenuation means 11b.
Control the amount of attenuation. Therefore, simultaneous transmission / reception (full-duplex) communication is not possible. However, in normal conversation, the speakers alternate, so this half-duplexity does not make conversation impossible. The digital signal processing is performed by a digital signal processing device (DSP).

Further, in the voice communication device having the hands-free function, the one side closed circuit is formed by the side-tone coupling generated on the transmission path and the acoustic coupling which occurs when the returned side-tone is output from the speaker 4 and wraps around the microphone 5. It is formed. Due to the acoustic coupling, the sidetone coupling, and the like, howling occurs when the gain (loop gain) of the closed loop becomes 1 or more. This howling does not occur if the echo cancellers 2 and 3 function ideally. However, the adaptive filters 13a, 13b
Due to the learning characteristic of, howling removal requires some convergence time. If the voice switch inserts an attenuation amount in the receiving path or the transmitting path, howling can be eliminated to some extent. As described above, various countermeasures have been proposed for detecting and removing howling.

[0006]

The attenuation amount control means 10 of the voice switch 1 shown in FIG. 6 compares the reception volume and the transmission volume as shown in FIG. 8 and determines whether the reception state or the transmission state. To do. The hysteresis as a criterion for judging the receiving state and the transmitting state has a width of about 20 dB. This is to prevent the voice switch 1 from very sensitively judging the transmitting state and the receiving state and switching them frequently. At point A, both the reception volume and the transmission volume are low, but normally, when the voice switch 1 is silent, it is determined that the voice switch 1 is in the reception state, and thus the reception state is determined. At point B, the reception volume is higher than the transmission volume, and the attenuation amount control means 10 determines that the reception state is in effect. At point C, the transmission volume is higher than the reception volume, so the attenuation amount control means 10 determines that it is in the transmission state. On the other hand, at the point D, the transmission volume and the reception volume are almost equal, and neither of them can be determined, and the voice switch 1 cannot switch to either the transmission state or the reception state.
The call becomes impossible. Here, from point B through point D to C
Considering the movement to the point, the phenomenon that the speaker's spoken word at the point C is cut off occurs. The same applies in the opposite case. That is, conversation cannot be performed immediately after switching the voice switch 1.

Further, when an object is placed on the voice communication equipment, the acoustic coupling field may fluctuate rapidly. At this time, a phenomenon called erroneous learning occurs in which the filter coefficient updating means 14a and 14b of the acoustic echo canceller 2 and the line echo canceller 3 update to erroneous filter coefficients. As a result, the balance between the line side and the sound side is significantly lost. That is, the gain of one of the line side and the acoustic side is 1 or more, and the other is 1 or less. In this state, the voice switch 1 is violently switched between the transmitting side and the receiving side. In this way, it becomes impossible to talk at all in the state where the switching is repeated.
This is called "blocking".

Blocking is difficult to distinguish from a hands-free state transition due to a change in voice energy in a normal conversation, because a completely divergent state does not continue. Therefore, blocking cannot be prevented by a conventional voice communication device having a hands-free function. On the other hand, digital mobile phones such as car phones have recently been required to be used in a hands-free state due to legal regulations. But,
When a digital mobile phone is put in a cup holder of an automobile and used, fluctuations in acoustic coupling are likely to occur and blocking is likely to occur.

The above problem can be solved by multiple filtering to completely eliminate the echo, but if the DSP is made to perform too complicated signal processing, the processing capacity of the DSP is significantly compressed. Becomes Furthermore, the power consumption of the DSP is also increased, which is a fatal problem for a digital mobile phone whose power consumption is limited. Therefore, it is desired to prevent blocking with the simplest processing possible. As a method for easily and easily preventing howling and blocking, the volume of a speaker or the like may be reduced, but the problem of inaudible sound is inevitable.

In view of the above problems, an object of the present invention is to easily and easily realize the stability and performance improvement of the hands-free function of a voice communication device.

[0011]

In order to achieve the above object, a first feature of the present invention is (a) an echo canceller for updating a filter coefficient according to an echo to generate a pseudo echo to remove the echo; (B) An audio switch that detects the power signal of the transmission / reception path and inserts an attenuation amount into either one of the transmission / reception path based on this power signal to switch between transmitting and receiving; In this case, the gist of the present invention is a hands-free communication system including at least a blocking detector that detects blocking based on the number of times the power signal and the voice switch are switched. Here, the “power signal” means a power component obtained from the reception signal and the transmission signal.

According to the hands-free communication system of the first feature of the present invention, by utilizing the property that the voice switch is violently switched many times when blocking occurs, the number of times of switching is counted to ensure the occurrence of blocking. Can be detected. That is, when switching between transmission and reception is repeated many times with a constant power, the blocking detector detects the occurrence of blocking.
When blocking is detected, the blocking can be removed by setting the attenuation amount to the maximum value and initializing the filter coefficient.

A second feature of the present invention is (a) a step of updating a filter coefficient according to an echo to generate a pseudo echo to remove the echo; (b) detecting a power signal of a transmission / reception path and detecting the power signal. Based on the step of switching the transmission and reception by inserting the attenuation amount into one of the transmission and reception paths; (c) When the voice switch is switched,
The gist is a blocking prevention method that includes at least a step of detecting blocking based on the number of times of switching of a power signal and a voice switch.

According to the blocking prevention method of the second feature of the present invention, the occurrence of blocking can be detected easily and easily without using a complicated method such as heavy use of filters.

A third feature of the present invention is (a) a command for updating a filter coefficient according to an echo to generate a pseudo echo to remove an echo; (b) detecting a power signal on a transmission / reception path and detecting the power signal. An instruction to switch the transmission and reception by inserting the attenuation amount in one of the transmission and reception paths based on
(C) A gist is a blocking prevention program including at least a command for detecting blocking based on a power signal and the number of times the voice switch is switched when the voice switch is switched.

According to the anti-blocking program of the third feature of the present invention, it can be executed by using a general-purpose DSP, so that the stability and the performance of all currently popular hands-free telephones can be improved. it can.

[0017]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings. In the first embodiment, a hands-free call system of a hands-free telephone will be described as an example. In the second embodiment, a hands-free call system for digital mobile phones will be described as an example. In the following description of the drawings, the same or similar parts are denoted by the same or similar symbols.

(First Embodiment) As shown in FIG. 1, a blocking prevention system according to a first embodiment of the present invention updates a filter coefficient according to an echo and generates a pseudo echo to generate an echo. Audio for canceling the sound side echo canceller 2, the line side echo canceller 3, and a voice signal for detecting the power signal of the transmission / reception path and inserting an attenuation amount in either one of the transmission / reception path based on the power signal to switch between transmission and reception. It has at least a switch 1 and a blocking detector 9 that detects blocking based on a power signal and the number of times the voice switch 1 is switched within a specified time. When detecting blocking, the blocking detector 8 sets the attenuation amount to the maximum value and initializes the filter coefficient.
Further, a howling detector 8 for detecting howling based on the power signal is provided, and when the howling detector 8 detects howling, the amount of attenuation is set to the maximum value and the filter coefficient is initialized. As shown in FIG. 2, the blocking detector 9 further performs blocking by a switching number counter 21 that counts the number of times the voice switch 1 is switched, a specified time timer 22 that sets a predetermined specified time, and a number of times the audio switch is switched during a specified time. Blocking judging means 23 for judging.

The voice switch 1 includes a first power detection filter 12a for detecting a power signal on the receiving path and a second power detection filter 12b for detecting a power signal on the transmitting path.
And a first attenuation means 11a for inserting an attenuation amount into the receiving path.
And a second attenuation means 11b for inserting an attenuation amount into the transmission path.
And the first and second damping means 11 based on the power signal.
It has a damping amount control means 10 which controls a and 11b and receives a command from the howling detector 8 and the blocking detector 9. During transmission, the line is dedicated for transmission when the strong voice amplitude is detected on the transmission side. Actually, this is because the received signal is attenuated by inserting an attenuation amount of about 40 dB (decibel) on the receiving side. On the other hand, when the transmission level becomes lower than a certain value, it is determined that the transmission is completed, and the signal line is dedicated to reception. Also in this case, 40d to the sender
A transmission signal is attenuated by inserting an attenuation amount of about B.
In either case, since the switch operation for stopping the operation of either the voice output or the voice input is performed, echo can be prevented.

The echo canceller is composed of echo cancellers 2 and 3 on the acoustic side and the line side, and the echo canceller 2 on the acoustic side receives the pseudo echo signal from the adaptive filter 13a which generates a pseudo echo signal from the received voice signal and transmits the voice. A first calculator 15a for subtracting the echo component of the signal,
And a filter coefficient updating means 14a for updating the filter coefficient of the adaptive filter 13a based on the echo component which is not erased by the calculator 15a. In the echo canceller 2 on the acoustic side, the pseudo echo is an output of the adaptive filter 13a on the acoustic side generated by a convolution operation based on the received signal and the filter coefficient of the adaptive filter 13a on the acoustic side. Also, on the line side, the echo canceller 3 on the line side is configured in the same manner. Therefore, the pseudo echo generated by the echo canceller 3 on the line side is the output of the adaptive filter 13b on the line side generated by the convolution operation based on the transmission signal and the filter coefficient of the adaptive filter 13b on the line side. Note that the filter coefficient updating means 31 does not update the filter coefficient when the attenuation amount control means 10 detects a so-called double talk when talking from both directions simultaneously. This is because if the filter coefficient is updated during double talk, the echo coefficient cannot be accurately canceled because the filter coefficient does not converge. Further, the attenuation amount of the voice switch 1 increases or decreases depending on the actual size of the echo. Therefore, when the echo cancellers 2 and 3 cancel the echo completely, the attenuation amount inserted by the voice switch 1 is reduced to about 6 dB.

Further, the output of the first echo canceller 2 is connected to the speaker 4 via the D / A converter 6.
An A / D converter 7 is connected to the input of the first echo canceller 2, and the microphone 5 is connected to the A / D converter 7.
Are connected.

Next, a blocking prevention method according to the first embodiment of the present invention will be described with reference to FIGS.

(A) First, in step S101 shown in FIG. 3, the received signal is filtered to produce noise.
To remove. Also, the external volume switch controls the reception volume set by the user. Next, in step S102, the transmission signal is filtered to remove noise. This filtering is D
This is performed by an interface such as the A / A converter 6 and the A / D converter 7. The external volume switch controls the transmission volume set by the user. It is the attenuation amount control means 10 that actually controls the audio volume.

(B) Next, in step S103, the calculation processing for updating the coefficient of the adaptive filter is performed by the filter coefficient updating means 14a and 14b. As the arithmetic processing, a learning identification method (N
LMS) is used. Taking the acoustic echo canceller as an example, the filter coefficient updating means 14a detects the acoustic echo component output from the transmitting side computing unit 15a. The filter coefficient updating means 14a is the adaptive filter 13 on the acoustic side.
a is updated to a filter coefficient suitable for minimizing the echo signal. By updating (learning) the filter coefficient, the pseudo echo generated by the acoustic adaptive filter 13a can be made closer to the original echo. Exactly the same processing is performed by the echo canceller 3 on the line side. The filter coefficient (tap coefficient) of the adaptive filter 13a on the acoustic side is
For example, the maximum number of taps can be 256. Recently, some have been expanded to 512 taps. on the other hand,
The filter coefficient of the line-side adaptive filter 13b is, for example, 3
2 taps. This is mainly due to the variability and the length of the generated echo. That is, acoustic coupling is more likely to occur than sidetone coupling, and the generated echo is long.

(C) Next, in step S104, the power signal of the received voice from the line is detected by the power detection filter 11a on the receiving side. At the same time, the power detection filter 11b on the transmission side detects the power signal in the transmission signal from the microphone 5. The detected power signals of the reception voice and the transmission voice are input to the attenuation amount control means 10.

(D) Next, in step S120, echo canceling processing for removing the echo component of the transmitted and received speech is performed. In step S105, the echo canceller 3 on the line side performs echo cancellation processing on the line side. The echo cancellation processing on the line side is performed by the adaptive filter 13b on the line side whose filter coefficient is updated in step S103. Adaptive filter 13b
Is an FIR filter, and a convolution operation is performed based on the transmission signal which is a digital signal and the filter coefficient of the adaptive filter 13b. As a result of the convolution operation, a pseudo echo signal is generated. The output of the adaptive filter 13b is input to the calculator 15b. The calculator 15b subtracts the echo component of the transmission signal including the echo signal from the pseudo echo signal to remove the echo. The residual echo component remaining without being removed is again detected by the filter coefficient updating means 14b. Next, also in step S106, similar processing is performed by the echo canceller 2 on the acoustic side.

(E) Next, in step S107, howling is detected based on the transmission voice power signal and the reception voice power signal input to the attenuation amount control means 10 in step S104. That is, the howling detector 8 is
It is determined from the transmitted voice power signal and the received voice power signal whether or not the gain of the closed loop is 1 or more. If the gain of the closed circuit is 1 or more, it is determined that howling is detected, and the process proceeds to step S110.

(F) In step S110, the attenuation amount of the voice switch 1 is set to the maximum value, and at the same time, the initialization processing of the adaptive filters 13a and 13b is performed. Here, the attenuation amount control means 10 sets the attenuation amount of the voice switch 1 to the maximum value. The maximum attenuation of the voice switch 1 is specifically about 40 dB. As a result, the howling can be eliminated by setting the gain of the closed loop to 1 or less. Also, the filter coefficient updating means 14a and 14b
The filter coefficients of are initialized. When howling is detected, the howling detector 8 outputs a voice switch attenuation amount maximizing and echo canceller initialization command.

(G) On the other hand, if it is determined in step S107 that the gain of the closed loop is less than 1 and there is no howling, the process proceeds to step S108. Step S108
Then, the blocking is detected by the processing as shown in FIG. First, in step S201 of FIG. 4, the switching number counter 21 shown in FIG. 2 determines whether or not the voice switch 1 has been switched. If the voice switch 1 is switched, the process proceeds to step S202. Step S
At 202, the switching number counter 21 determines whether or not the power signals detected by the power detection filters 12a and 12b shown in FIG. 1 have a certain magnitude. This is because a large power signal is generated in the transmission channel and the reception channel when blocking occurs. If the detected power signal is greater than or equal to the threshold value, the process proceeds to step S203. In step S203, the switching number counter 21 shown in FIG. 2 counts the transmission / reception switching number of the voice switch 1. The specified time timer 22 shown in FIG. 2 sets a specified specified time.
Therefore, it is possible to calculate the number of transmission / reception switching times within the specified time of the voice switch 1 shown in FIG. After calculating the number of transmission / reception switching within the specified time, the process proceeds to step S204 shown in FIG. In step S204, the blocking determination means 23 shown in FIG. 2 is used to determine whether or not the number of switchings within a specified time exceeds a predetermined threshold value. If it exceeds the threshold, the process proceeds to step S205 shown in FIG.
After that, the process proceeds to step S110 in FIG. Step S11
At 0, the attenuation amount of the voice switch is set to the maximum value, and at the same time, the initialization processing of the adaptive filters 13a and 13b is performed. When blocking is detected, the blocking detector 9 outputs a voice switch attenuation amount maximizing and echo canceller initialization command.

(H) On the other hand, if the voice switch is not switched in step S201, step S20
If the power signal is less than the threshold value in step 2, step S
If the number of switchings within the specified time is less than the predetermined threshold value in 204, it is determined that blocking is not detected, and the process proceeds to step S206, and then proceeds to step S109 shown in FIG. Since neither howling nor blocking occurs in step S109, the attenuation amount control means 10 shown in FIG. 1 calculates and controls the attenuation amount of the audio switch 1 based on the power signal detected in step S104. After that, the process proceeds to step S111.

(I) In step S111, either the attenuation means 11a or 11b inserts the attenuation amount based on the instruction of the attenuation amount control means 10 and the voice switch process is performed.
When howling and blocking occur, the maximum attenuation amount is set in the attenuation amount control means 10. When howling and blocking have not occurred, the attenuation amount control means 10 is set with an attenuation amount proportional to the power signals of the transmitted and received signals. After performing the voice switch process, the process is restarted from step S101. First
According to the embodiment, even if blocking occurs, it can be detected immediately. Further, after the blocking is detected, the blocking can be removed by maximizing the attenuation amount. In addition, the adaptive filter 13a,
By initializing 13b, blocking due to erroneous learning can also be prevented. Also, both howling and blocking can be detected and removed. Further, since the unstable state can be immediately restored, a normal conversation can be continued even if blocking occurs. Since it can be implemented using a general-purpose DSP, it can be applied to any digital voice communication system having a hands-free function (second embodiment), as shown in FIG. The hands-free call system for digital mobile phones according to the embodiment differs from the hands-free call system shown in FIG. 1 in that an echo canceller on the line side is not provided. This is because, in the digital mobile phone, the echo canceller on the line side is provided in the telephone station (base station). Further, a reception signal register 35 is further provided between the reception path and the adaptive filter 32. A filter coefficient register 36 is further provided between the filter coefficient updating means 31 and the adaptive filter 32. This solves the delay of the received signal due to the decoding of the voice signal. still,
Illustration of the D / A converter 6 and the A / D converter 7 shown in FIG. 1 is omitted.

Upon reception, the reception signal transmitted from the telephone station is input to the codec 95 via the interface (I / F) 51 on the telephone station side. Codec 95
It includes a channel codec (Ch-CODEC) 52 and a speech codec (Sp-CODEC) 53.
The channel codec (Ch-CODEC) 52 performs error correction decoding processing. Speech codec (Sp-C
In ODEC) 53, audio decoding processing is further performed. The noise cancellers (NC) 59a and 59b remove background noise included in the transmitted and received signals. Equalizer (EQ) 57
Is a variable filter that adjusts the frequency characteristic of the received signal. Further, it works to improve the feeling of hearing. The volume control 58 is -25 according to the command of the volume switch operated by the user.
The volume is adjusted within the range of dB to +18 dB. However, when the volume is increased, howling is more likely to occur, and therefore the amount of attenuation inserted by the voice switch 33 is increased or decreased in conjunction with each other.

On the other hand, at the time of transmission, the transmission signal is subjected to voice coding processing by the speech codec (Sp-CODEC) 53. The channel codec (Ch-CODEC) 52 further performs correction coding processing. Then, it is transmitted to the telephone station (base station) via the interface (I / F) 51 on the telephone station side. The encoding method is full rate (9
The VSELP (Vector Sum Excitation Linear Prediction) method is used at 600 bps, and the PSI-CELP method is used at the half rate (4800 bps). As described above, in the digital automobile / mobile phone system, in order to effectively use the frequency, the device for transmitting the information is reduced as much as possible. Also, the first switch SW1, the second switch SW2, and the VOX means 5 provided in the transmission path.
5 turns on only when an input voice is detected and voice is present.
On the other hand, it turns off when there is no sound. The third switch SW3 provided on the receiving path is used for intermittent reception. That is,
The transmitted signal includes information to be received for each predetermined time slot. Third switch S
W3 is turned on / off according to this time slot. As a result, it is possible to reduce power consumption during transmission and reception. It also solves audible problems caused by handover (changing base stations) and silence compression processing.

Next, a blocking prevention method according to the second embodiment of the present invention will be described with reference to FIGS. However, a part of the description overlapping with the blocking prevention method according to the first embodiment of the present invention will be omitted. (A) First, in step S301 shown in FIG.
The noise canceller (NC) 59a removes noise from the received signal decoded by the codec 53 shown in FIG. The received signal from which noise has been removed is subjected to frequency characteristic adjustment, that is, phase compensation, by an equalizer (EQ) 57. Equalizer (E
The volume of the received signal whose phase is compensated by Q) 57 is adjusted by the volume control 58 to the value set by the user with the volume switch. Next, in step S302 shown in FIG. 6, the noise canceller 59b removes noise in the transmission signal. The sound volume is also adjusted by the attenuation amount control means 10 of the sound switch 33.

(B) Next, in step S303, the calculation processing for updating the coefficient of the filter coefficient register 36 is performed by the filter coefficient updating means 31 shown in FIG. As the arithmetic processing, a learning identification method (NLMS) which is a normalized least square method (LMS) is used. However, since the reception signal of the mobile terminal is decoded through the codec 95, the reception signal is slightly delayed. Therefore, it is necessary to match the input timings of the filter coefficient updating means 31 and the received signal to the adaptive filter 32. Filter coefficient updating means 3
The filter coefficient updated by 1 is once stored in the filter coefficient register 36 and input to the adaptive filter 32. Further, the reception signal is temporarily stored in the reception signal register 35,
It is input to the adaptive filter 32.

(C) Next, in step S304, the power signal of the receiving side power detection filter 12a shown in FIG. 5 detects the power signal of the receiving voice from the line. At the same time, the transmission side power detection filter 12b detects the power signal of the transmission voice from the microphone 5. The power signals of the received voice and the transmitted voice detected are the attenuation amount control means 1 shown in FIG.
Input to 0.

(D) Next, in step S305, echo canceling processing for removing the echo component of the transmitted signal is performed. In the reception signal register 35 shown in FIG. 5, the digital reception signal which is the output of the volume control 58 is fetched and accumulated at a constant sample timing. Next, in the adaptive filter 32, a convolution operation is performed based on the digital received signal accumulated in the reception signal register 35 and the filter coefficient stored in the filter coefficient register 36. As a result, a pseudo echo is generated. This pseudo echo signal is sent to the calculator 37 to remove the echo component in the transmission signal.

(E) Next, in step S306, the howling detector 8 makes a loop circuit based on the transmitted voice power signal and the received voice power signal input to the attenuation amount control means 10 shown in FIG. 5 in step S304. Howling is detected by judging that the gain is 1 or more. If howling is detected, the process proceeds to step S309. In step S309, the maximum attenuation amount of the voice switch 33 and the initialization process of the adaptive filter 32 shown in FIG. 5 are performed.
However, since the filter coefficient is stored in the filter coefficient register 36, the filter coefficient register 36 is actually initialized.

On the other hand, if howling is not detected in step S306, the process proceeds to step S307.
In step S307, blocking is detected by the processing shown in FIG. First, step S201 in FIG.
2, the switching counter 21 shown in FIG. 2 determines whether or not the voice switch 33 shown in FIG. 5 has been switched. If the voice switch 33 has been switched, the process proceeds to step S202. In step S202, the switching number counter 21 determines whether or not the power signals detected by the power detection filters 12a and 12b shown in FIG. 5 have a certain magnitude. This is because a large power signal is generated in the transmission channel and the reception channel when blocking occurs. If the detected power signal is greater than or equal to the threshold value, the process proceeds to step S203. In step S203, the switching number counter 21 shown in FIG. 2 causes the voice switch 33 shown in FIG.
It counts the number of times that the voice transmission and reception are switched. The specified time timer 22 shown in FIG. 2 sets a specified specified time. Therefore, it is possible to calculate the number of transmission / reception switching times within the specified time of the voice switch 33. After calculating the number of transmission / reception switching within the specified time, the process proceeds to step S204 shown in FIG. In step S204, the blocking determination means 23 shown in FIG. 2 is used to determine whether or not the number of switchings within a specified time exceeds a predetermined threshold value. If it exceeds the threshold, the process proceeds to step S205 shown in FIG. 4, and then proceeds to step S309 of FIG. In step S309, FIG.
The attenuation of the voice switch 33 shown in is set to the maximum value,
At the same time, the initialization process of the adaptive filter 32 is performed. still,
When blocking is detected, the blocking detector 9 outputs a command for maximizing the attenuation amount of the voice switch 33 and initializing the echo canceller 34. In step S309,
Voice switch 33 Attenuation maximization, echo canceller 34
After the initialization, the process proceeds to step S310.

(G) On the other hand, in step S201, when the voice switch 33 shown in FIG. 5 is not switched,
If the power signal does not reach the threshold value in step S202, or if the number of times of switching within the specified time does not reach the predetermined threshold value in step S204, it is determined that blocking has not been detected, and the process proceeds to step S206.
Go to step S308 shown in. Since neither howling nor blocking has occurred in step S308, based on the power signal detected in step S304, as shown in FIG.
The attenuation amount control means 10 shown in (1) calculates and controls the attenuation amount of the voice switch 33. Then, it progresses to step S310.

(H) In step S310, either the receiving side attenuating means 11a or the transmitting side attenuating means 11b inserts the amount of attenuation based on the instruction from the amount of attenuation controlling means 10 shown in FIG. Perform switch processing. When howling and blocking occur, the attenuation amount control means 10
Is set to the maximum attenuation. If howling and blocking do not occur, the attenuation amount control means 1
In 0, the amount of attenuation proportional to the power signals of the transmitted and received signals is set. After performing the voice switch process,
The process is restarted from step S301.

According to the second embodiment of the present invention, it is possible to minimize the occurrence of blocking even in a digital mobile phone by simply modifying the program. Therefore, the processing capacity of the DSP is not compressed. As a result, a stable hands-free call system can be realized without increasing the processing time and power consumption of the DSP.

(Other Embodiments) Although the present invention has been described by the above embodiments, it should not be understood that the description and drawings forming a part of this disclosure limit the present invention. From this disclosure, various alternative embodiments, examples and operational techniques will be apparent to those skilled in the art.

As described above, when blocking occurs, conversation cannot be performed at all. Therefore, it is necessary to detect blocking earlier. In the above-mentioned first and second embodiments, the blocking detector 9 detects blocking after the howling detector 8 detects howling.
Of course, the blocking detector 9 may first detect the blocking. In this case, in the flowchart shown in FIG. 3, the content of the process of step S107 and the content of the process of step S108 may be exchanged.

Further, in the first and second embodiments,
The howling detector 8 and the blocking detector 9 in the DSP output an attenuation amount maximizing instruction to the attenuation amount controlling means 10 and a filter coefficient initializing instruction to the filter coefficient updating means 31, but a main CPU (not shown) (central processing) The control device) may output this command from the outside.
Further, the hands-free telephone and the digital mobile telephone have been described as an example, but it goes without saying that the blocking prevention method of the present invention can be applied to various digital communication devices such as a video telephone and a FAX.

As described above, it goes without saying that the present invention includes various embodiments not described here. Therefore, the technical scope of the present invention is defined only by the matters specifying the invention according to the scope of claims reasonable from the above description.

[0047]

According to the present invention, the stability and performance of the hands-free function of a voice communication device can be easily and easily realized.

According to the present invention, it is possible to provide a blocking prevention method capable of surely removing blocking without causing a reduction in the processing capacity of the DSP and an increase in power consumption.

Further, according to the present invention, since it can be implemented using a general-purpose DSP, it is possible to provide a blocking prevention program that can be installed in any digital voice communication device having a hands-free function.

[Brief description of drawings]

FIG. 1 is a block diagram of a handsfree call system for a handsfree telephone according to a first embodiment of the present invention.

FIG. 2 is a block diagram of a blocking detector according to the first embodiment of the present invention.

FIG. 3 is a flowchart showing a blocking prevention method according to the first embodiment of the present invention.

FIG. 4 is a flowchart showing processing of a blocking detector according to the first exemplary embodiment of the present invention.

FIG. 5 is a block diagram of a handsfree call system for a digital mobile phone according to a second embodiment of the present invention.

FIG. 6 is a flowchart showing a blocking prevention method according to a second embodiment of the present invention.

FIG. 7 is a block diagram of a conventional hands-free call system.

FIG. 8 is a graph showing the operation of a conventional voice switch.

[Explanation of symbols]

1 voice switch 2 Acoustic echo canceller 3 Line side echo canceller 4 speakers 5 microphones 6 D / A converter 7 A / D converter 8 Howling detector 9 Blocking detector 10 Attenuation control means 11a First (listening side) attenuation means 11b Second (sending side) attenuation means 12a First (receiver side) power detection filter 12b Second (transmitting side) power detection filter 13a First (sound side) adaptive filter 13b Second (line side) adaptive filter 14a First (sound side) filter coefficient updating means 14b Second (line side) filter coefficient updating means 15a First (sending side) computing unit 15b Second (receiver side) computing unit 21 Switching counter 22 Specified time timer 23 Blocking determination means 31 filter coefficient updating means 32 adaptive filters 33 voice switch 34 Echo canceller 35 Receive signal register 36 Filter coefficient register 37 calculator 51 Line side interface 52 channel codec 53 speech codecs 54 Sound side interface 55 VOX means 59a, 59b Noise canceller 57 Equalizer 58 Volume control 91, 92, 93 DSP 95 codecs

Claims (12)

[Claims] 1. An echo canceller that updates a filter coefficient according to an echo to generate a pseudo echo to remove the echo, and a power signal of a transmission / reception path and detects either one of the transmission / reception path based on the power signal. A voice switch for switching between transmission and reception by inserting an attenuation amount into the voice switch, and a blocking detector that detects blocking based on the number of times the power signal and the voice switch are switched when the voice switch is switched, A hands-free call system having at least the following. 2. The blocking detector sets the attenuation amount to a maximum value and initializes the filter coefficient when blocking is detected.
Hands-free calling system described. 3. A howling detector for detecting howling based on the power signal, wherein the howling detector sets the attenuation amount to a maximum value when the howling is detected, and initializes the filter coefficient. The hands-free call system according to claim 2. 4. The blocking detector determines the occurrence of blocking from the switching count counter that counts the number of times the voice switch is switched, a specified time timer that sets a specified time, and the number of times of switching within the specified time. 4. The hands-free call system according to claim 3, further comprising blocking determination means for performing the blocking determination. 5. A step of updating a filter coefficient according to an echo to generate a pseudo echo to remove the echo, and detecting a power signal of a transmission / reception path and applying the power signal to one of the transmission / reception paths. It includes at least a step of inserting an attenuation amount to switch between transmitting and receiving, and detecting blocking based on the number of times of switching of the power signal and the voice switch when the voice switch is switched. A method for preventing blocking, which comprises: 6. The blocking prevention method according to claim 5, further comprising the step of maximizing the amount of attenuation and initializing the filter coefficient when the blocking is detected. 7. The method further comprises the step of detecting howling based on the power signal, wherein the step of detecting the howling maximizes the attenuation amount when the howling is detected and initializes the filter coefficient. 7. The blocking prevention method according to claim 6. 8. The step of detecting the blocking, the step of determining whether the voice switch has been switched, the step of determining whether the power signal exceeds a predetermined threshold value, 8. The blocking prevention method according to claim 6, further comprising the step of determining whether or not the number of times the voice switch is switched exceeds a threshold value. 9. An instruction for updating a filter coefficient according to an echo to generate a pseudo echo to remove the echo, and a power signal of a transmission / reception path is detected, and one of the transmission / reception paths is detected based on the power signal. It has at least a command for switching between transmission and reception by inserting an attenuation amount, and a command for detecting blocking based on the number of switching times of the power signal and the voice switch when the voice switch is switched. A blocking prevention program characterized by the following. 10. The anti-blocking program according to claim 9, further comprising an instruction for maximizing the attenuation amount and initializing the filter coefficient when the blocking is detected. 11. An instruction for detecting howling based on the power signal is further included, wherein the instruction for detecting howling maximizes the attenuation amount when detecting howling and initializes the filter coefficient. The blocking prevention program according to claim 10. 12. The command for detecting blocking is a command for determining whether or not the voice switch has been switched, a command for determining whether or not the power signal is equal to or more than a predetermined threshold, and the command within a predetermined time. The blocking prevention program according to claim 10 or 11, comprising an instruction for determining whether or not the number of times the voice switch is switched exceeds a threshold value.