JPH0691576B2 - Loud phone - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a loudspeaker telephone suitable for use in a car telephone, a conference telephone, etc., and particularly, a microphone in which sound from a speaker is reflected on a wall of a room or the like. The present invention relates to an echo-cancell type loudspeaker system which is capable of preventing howling due to acoustic coupling.

[Conventional technology]

The loudspeaker telephone is not provided with a handset (handset), and is capable of making a call using a fixed speaker and a microphone (hereinafter referred to as a microphone). A transmission voice signal generated by inputting a transmission voice to a microphone is amplified by a transmission amplifier, and then a 2-line-
The received voice signal is transmitted through a hybrid circuit that performs 4-wire conversion, and the received voice signal passes through the hybrid circuit, is amplified by a receiving amplifier, and is then supplied to a speaker to generate a received voice. Therefore, the loudspeaker phone allows both hands to be used freely during a call, and the car phone, which has been attracting attention in recent years, allows safe driving while talking.
All of the conference telephones are useful in that they can be easily written while talking.

By the way, when the public address telephone is used for a car phone or a conference call, the public address telephone is installed in a car or a room. For this reason, the received sound output from the speaker is reflected by the window glass in the vehicle or the wall of the room, is input to the microphone, and the output signal is reflected by the hybrid circuit and is supplied to the speaker. become. Therefore, if a loudspeaker telephone is simply composed of a microphone, a speaker, and an amplifier,
The loop of this signal may cause howling, which may make the call impossible.

Conventionally, in order to prevent this howling, a voice switch system has been used in a loudspeaker telephone. This is a system for breaking this loop by inserting a means which causes a loss in the receiving signal path at the time of transmitting and, conversely, at the transmitting signal path at the time of receiving. Therefore, in this method, a loss occurs in any one of the communication paths, and simultaneous communication is impossible. In addition, there is a drawback that disconnection of the beginning and end of the speech occurs due to loss switching, giving an unnatural feeling to the speech.

As a substitute for this method, the echo cancel method has attracted attention in recent years.

As shown in FIG. 7, this echo-cancelled loudspeaker telephone has a signal s (t) + y (y) input to the microphone 1.
Among them, the echo canceling circuit 3 for canceling only the signal y (t) output from the speaker 2 and reflected by the wall of the room or the like is provided.

The transmission signal s (t) is input to the microphone 1, passes through the echo canceling circuit 3, is amplified by the transmission amplifier 4, and is then transmitted from the hybrid circuit 6 to the subscriber line 7 to be transmitted. Also, it is sent to the subscriber line 7 and transmitted. Further, the reception signal x (t) sent through the subscriber line 7 is
The signal is supplied from the hybrid circuit 6 to the reception amplifier 5, amplified there, and then supplied to the speaker 2.

In the echo canceling circuit 3, the transversal filter 31
Is a filter having a characteristic that approximates a transfer function (this is called an echo path) in which an input signal x (t) to the speaker 1 becomes y (t) due to reflection by a wall of a room (this is called a pseudo echo path). ), And is generally a filter having an impulse response of the transfer function and an approximate tap coefficient. The transversal filter 31 produces a pseudo echo signal y (t) from the signal x (t), and the input signal y (t) + s of the microphone 1 is generated.
The subtracter 32 subtracts the pseudo echo signal (t) (≈y from (t)).
By subtracting (t)), only the echo signal y (t) is canceled and the transmission signal s (t) is output.

The tap coefficient of the transversal filter 31 is the input signal x (t) to the speaker 2 and the output signal e of the subtractor 32.
From (t), the LMS method (Least Mean Square Method) or the learning identification method and the well-known algorithm,
It is sequentially obtained by the tap coefficient estimating circuit 33. The tap coefficient estimation circuit 33 uses this algorithm to input the signal e
The tap coefficient is successively corrected based on (t) and the input signal x (t) to the speaker 2, and the tap coefficient is sequentially given to the transversal filter 31 as the tap coefficient. Finally, the tap coefficient of the transversal filter 31 is approximated to the impulse response of the echo path, and the echo signal y (t) is converted from the input signal s (t) + y (t) of the microphone 1.
Almost cancel.

It is necessary to estimate the tap coefficient when the input signal to the microphone 1 is only the echo signal y (t).
Otherwise, the echo signal y is transmitted by the transmitted signal s (t).
(T) is masked and tap coefficient estimation becomes inaccurate. Therefore, the input signal s (t) + y (t) of the microphone 1
And the input signal to the speaker 2, that is, the reception signal x (t)
, And the input signal S (t) + y of the microphone 1 is compared.
When the power of (t) is larger than the power of the reception signal x (t) by a certain amount, it is determined that the transmission signal s (t) is present, and the double talk for prohibiting the tap coefficient of the tap coefficient estimation circuit 33 from being updated. A detector 34 is provided.

In this way, the loop of the previous signal is broken between the microphone 1 and the speaker 2, and howling is prevented. Also,
With this method, unlike the voice switch method, it is not necessary to insert a loss in the speech path, so that simultaneous speech can be made and good speech quality can be obtained without disconnecting the beginning and ending of words.

By the way, in order for the transversal filter to closely approximate the echo path, that is, in order to sufficiently cancel the echo signal y (t) from the input signal of the microphone, it takes some time. This is usually started by setting the tap coefficient value to zero at the beginning, updating the tap coefficient sequentially according to the previous algorithm, and finally approximating the impulse response of the echo path. It depends on the number of tap coefficients,
It takes several 100 msec to several sec. During this period, sufficient echo cancellation is not performed (the amount of cancellation is small), so howling may occur.

In order to prevent this, conventionally, as described in JP-A-58-90832, JP-A-60-117928, JP-A-61-3536, etc., noise and a chirp sequence signal prior to a call are disclosed. , Sweep sine wave signal or impulse sequence signal is generated from the speaker as a training signal, tap coefficient is calculated in advance by the previous algorithm, and the echo canceling circuit operation is started with the tap coefficient to talk. Things are known.

[Problems to be solved by the invention]

However, in the above-mentioned related art, since a special training signal is used, the sound generated thereby gives the user an unpleasant feeling. Moreover, since this is always generated prior to a call, it further annoys the user.

Further, the timing of sending these training signals is out of the ordinary telephone operation sequence, which gives the user an unnatural feeling.

An object of the present invention is to solve the above problems and provide a loudspeaker telephone capable of sufficiently removing howling without causing unnaturalness or discomfort to the user.

[Means for solving problems]

In order to achieve the above object, the present invention provides a period of an audible sound such as a bell sound and a generated sound that is generated prior to transmission and reception.
A switch that opens the transmission signal path and closes the transmission signal path during a call is provided to operate the echo canceling circuit during the audible sound period and the communication period.

[Action]

Since the transmission signal path is opened during the audible sound, the audible sound emitted from the speaker is reflected by the wall of the room and input into the microphone, and the output signal of this microphone is sent only to the echo canceller. Therefore, the echo canceling circuit can detect the characteristic of the echo path, that is, learn the echo path by the audible sound. With the learning result as initial data, the echo canceling circuit learns the echo path during a call and cancels the echo signal from the output signal of the microphone.
Therefore, the echo signal is sufficiently and correctly erased from the start of the call, and a good call is possible, and the audible sound is a voice that the user normally experiences when using the phone, and an unpleasant sound is generated. It never happens.

〔Example〕

 Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing an embodiment of a loudspeaker telephone according to the present invention, in which 8 is a switch, 9 is a dial tone generating circuit, and 10 is a circuit.
Is an automatic fax switch, 11 is a calling signal detection circuit, 12
Is a bell sound generation circuit, 13 is a ringing tone detection circuit, 14 is a busy tone detection circuit, 15 is a mixing circuit, 16 is a tap coefficient holding circuit,
17 is a push button switch, 18 is a push button switch, 19
Is a control circuit, and the parts corresponding to those in FIG.

In the figure, a switch 8 opens and closes between the transmitting amplifier 4 and the hybrid circuit 6. The dial tone generation circuit 9 generates an audible tone such as a DTMF signal in order to notify the exchange (not shown) of the telephone number of the other party at the time of calling. The automatic fax switch 10 connects the hybrid circuit 6 to the subscriber line 7. The call signal detection circuit 11 detects a call signal from the exchange. The bell sound generation circuit 12 is driven by the output signal of the call signal detection circuit 11 and generates an audible sound (bell sound) that informs the user that there is a call (incoming call). The ringing tone detection circuit 13 detects the ringing tone sent by the exchange to inform the caller that the dialed party is being called. The busy tone detection circuit 14 detects a busy tone sent by the exchange to notify the caller that the dialed party is busy. Mixing circuit 15
Electrically mixes the signal from the hybrid circuit 6 and the output signal of the bell sound generating circuit 12. The tap coefficient holding circuit 16 temporarily holds the tap coefficient data of the transversal filter 31 which is the output of the tap coefficient estimating circuit 33. The push button switch 17 is for notifying the voice telephone of the user's call intention, and the push button 18 is for inputting the telephone number of the other party. The control circuit 19 controls the whole.

Next, the operation of this embodiment will be described. First, the operation at the time of incoming call will be described with reference to FIG.

When a call signal is sent from the exchange (not shown) through the subscriber line 7, the call signal detection circuit 11 detects the call signal and notifies the control circuit 19 that a call has been made (step 201). Therefore, the control circuit 19
At the same time that the switch 8 is opened (step 202), the bell sound generation circuit 12 is controlled and the bell sound signal is generated by the output signal of the calling signal detection circuit 11. This bell sound signal is amplified by the reception amplifier 5 via the mixing circuit 15 and is emitted by the speaker 2. The user can know that there is an incoming call with this bell sound. Further, the control circuit 19 operates the echo canceling circuit 3 (above, step 203).

The user hears the bell and, if he / she wants to make a call, presses the hook button switch 17 (step 204). Control circuit 19
When the push button switch 17 is pressed, the tap coefficient data output from the tap coefficient estimating circuit 33 at that time is stored in the tap coefficient holding circuit 16 (step 205), and the echo canceling circuit 3 operates. Stop (step 206). Then, the bell sound generation circuit 12 is controlled to stop the sounding of the bell sound (step 207).

In the above operation, while the bell sound is being emitted by the bell sound generation circuit 12, the bell sound is reflected by the wall of the room or the like and input to the microphone 1, and the output signal of the microphone 1 is the echo canceling circuit. 3 is supplied. Therefore, the echo canceling circuit 3 uses the bell sound to perform the echo canceling operation described with reference to FIG. 7 according to a predetermined algorithm to obtain the characteristics of the echo path of the room, that is, the impulse response of the room. The tap coefficient estimation circuit 33 sequentially finds. Therefore, when the hook button switch 17 is pressed, the tap coefficient data which sufficiently represents the echo path characteristics of the room at that time is stored in the tap coefficient holding circuit 16. At this time, the switch 8 is open, and since the signal loop is not formed, the howling is not exceeded.

When the operation of the bell sound generation circuit 12 is stopped, the control circuit 19 supplies the tap coefficient data stored in the tap coefficient holding circuit 16 to the transversal filter 31 as an initial value, and the echo canceling circuit 3 is operated again ( Step 20
8). Then, the switch 8 is closed (step 209), and then the automatic fax switch 10 is closed (step 210).
The hybrid circuit 6 is connected to the subscriber line 7. This allows the user to talk to the other party (step
211). At this time, since the echo canceling circuit 3 starts its operation from a state having a sufficient echo canceling characteristic, it hardly causes howling even if a signal loop is formed, and a comfortable call can be performed from the start. it can. When the user presses the fax switch button 17 again (step 212), when the call ends, the control circuit 19 stops the operation of the echo canceling circuit 3 (step 213) and opens the automatic fax switch 10 (step 214). The hybrid circuit 6 is disconnected from the subscriber line 7.

Next, the operation at the time of calling will be described with reference to FIG.

Instead of lifting the handset on a regular phone,
Press the hook button switch 17 (step 301) and enter the telephone number from the push button 18 to make a call. When the control circuit 19 detects that the hook button switch 17 is pressed, it opens the switch 8 (step 302) and starts the operation of the echo canceling circuit 3 (step 30).
3). Then, the automatic fax switch 10 is closed (step 304) and the hybrid circuit 6 is connected to the subscriber line 7.

When the user inputs the telephone number of the other party by using the push button 18 (step 305), the control circuit 19 detects this telephone number and controls the dial tone generation circuit 9 to make two different numbers for each telephone number. The hybrid circuit 6 generates a two-frequency signal (DTMF signal) obtained by mixing frequency signals.
Sent to the subscriber line 7 and transmitted to the exchange. This DT
The MF signal is also bypassed by the hybrid circuit 6 and emitted by the speaker 2 through the mixing circuit 15 and the reception amplifier 5 (step 306). The user can confirm that he / she has input the telephone number by this vocalization.

Further, the sound from the speaker 2 is reflected by the wall of the room or the like and input to the microphone 1, and the output signal of the microphone is supplied to the echo canceling circuit 3. In the echo canceling circuit 3, the impulse response of the room is sequentially estimated from the signal supplied from the dial tone generating circuit 9 through the hybrid circuit 6, the mixing circuit 15 and the receiving amplifier 5 and the signal from the microphone 1. Required by circuit 33.

When the input of the telephone number is completed, the exchange connects the line to the other party. At this time, if the other party is in a call, a busy tone signal is sent to the calling terminal (loudspeaker telephone), and if not, a calling signal is sent to the calling party and a calling tone signal is sent to the calling terminal.

If the other party is talking, the busy sound detection circuit 14 detects this and notifies the control circuit 19 (step 307). Also,
This busy sound passes through the hybrid circuit 6 and the mixing circuit 15,
The sound is emitted from the speaker 2 via the receiving amplifier 5 (step 309). The user hears this and knows that the other party is talking, and presses the hook button 17 again (step 31
0). This is the same operation as placing a handset on a normal telephone. As a result, the control circuit 19 causes the echo canceling circuit 3
Is stopped (step 319), the automatic fax switch 10 is opened (step 320), and the hybrid circuit 6 is separated from the subscriber line 7.

If the other party is not in a call, the ringing tone detection circuit 13 detects the ringing tone signal from the exchange and informs the control circuit 19 of the detection result (step 308). Further, the ringing tone signal passes through the hybrid circuit 6 and the mixing circuit 15,
It is supplied to the speaker 2 through the receiving amplifier 5 and the ringing tone is emitted from this (step 311). The user
By listening to this ringing tone, it is possible to know that the exchange is calling the other party. The exchange stops sending the ringing tone signal when the other party lifts the handset of the telephone for a call. When the ringing tone detection circuit 13 detects the stoppage of the ringing tone signal (step 312), the ringing tone from the speaker 2 disappears, whereby the user knows that the other party answered the telephone. You can

When the ringing tone detection circuit 13 detects the stop of the ringing tone, the control circuit 19 stores the tap coefficient of the tap coefficient estimating circuit 33 at the time of this stop in the tap coefficient holding circuit 16 (step 314). And echo canceling circuit 3
Stop the operation of (step 314).

Then, the tap coefficient data stored in the tap coefficient holding circuit 16 is used as an initial value for the transversal filter.
31 to reactivate the echo canceling circuit 3 (step 31
5) Then, switch 8 is closed (step 316). This allows the user to talk to the other party (step 31
7). At this time, the echo canceling circuit 3 sufficiently learns the echo path characteristics of the room using the dial tone and the ring tone, and starts operation with the tap coefficient as an initial value, so that a loop of signals is formed and howling is caused. You can make comfortable calls without making a call.

When the user presses the hook button switch 17 again (step 318) and the call ends, the control circuit 19 stops the operation of the echo canceling circuit 3 (step 319), and the automatic fax switch 10 is opened (step 320). The circuit 6 is disconnected from the subscriber line 7.

As described above, in this embodiment, the echo canceling circuit 3 is preliminarily set after the signal loop is cut off using the audible sound (bell sound at the incoming call, dial tone and ringing sound at the outgoing call) necessary for the telephone. It is operated, the impulse response of the room is sequentially obtained and stored by a predetermined algorithm, and when a call is entered,
Since the echo canceling operation is performed by closing the signal loop and using the impulse response obtained earlier as the initial value, a sufficient echo canceling amount can be obtained from the beginning during a call, and a comfortable call without howling can be started. It is possible from

In this embodiment, the filter simulating the echo path is a transversal filter, but the filter is not limited to this, and another filter such as a recursive lattice filter may be used.

Further, in the above operation, the operation of the echo canceling circuit 3 is temporarily stopped when the audible sound ends (steps 206 and 314),
Reactivated when the call started (steps 208,315)
However, it may be operated continuously without being stopped. In this case, the tap coefficient holding circuit 16 does not need to hold the tap coefficient. Further, the learning at the time of calling may be performed only by the dial tone.

FIG. 4 is a block diagram showing another embodiment of the loudspeaker telephone according to the present invention, in which reference numeral 20 is a selection switch, portions corresponding to those in FIG.

In FIG. 4, the selection switch 20 is controlled by the control circuit 19, and as one input of the mixing circuit 15, the output of the dial tone circuit 9 (A side) or the output of the hybrid circuit 6 (B side).
Select.

In the embodiment shown in FIG. 1, since the dial tone signal is bypassed by the hybrid circuit 6 and supplied to the mixing circuit 15, its bell is lowered by about 10 dB and is uttered. The selection switch 20 prevents this, and while learning the echo path of the room by using the dial tone, the selection switch 20 is connected to the B side, that is, the dial tone signal output by the dial tone generating circuit 9. Is directly supplied to the mixing circuit 15 to prevent its level from being lowered. The other operations are similar to those of the embodiment shown in FIG. 1, and therefore further description will be omitted.

FIG. 5 is a block diagram showing still another embodiment of the loudspeaker according to the present invention, in which 21 is a non-linear circuit, 22 is a selection switch, and the parts corresponding to those in FIG. The description will be omitted.

In FIG. 5, the selection switch 22 is closed to the B side during a call and is closed to the A side while learning the echo path of the room by using an audible sound. The bell sound, dial tone, and ringing sound are distorted by the non-linear circuit 21, exchanged for a signal containing a lot of high and low harmonic components, and then emitted from the speaker 2. In each of the embodiments described above, the audible sound is close to a sine wave, and the number of frequency components is small. Therefore, if the impulsive response of the room is estimated with the waveform of the audible sound as it is, the estimation accuracy may be deteriorated because it is significantly different from the waveform during the call. The non-linear circuit 21 is provided to prevent this, and is to generate distortion by performing processing such as clipping and slicing on the audible sound signal. Therefore, the sound emitted from the speaker 2 includes many frequency components, and the accuracy of estimating the impulse response of the room is improved.

FIG. 6 is a block diagram showing still another embodiment of the loudspeaker telephone according to the present invention, in which 23 is a voice synthesizing circuit, the parts corresponding to those in FIG. To do.

As described in the embodiment of FIG. 5, the bell sound, the dial tone, and the ring tone do not consist of many frequency components,
The chamber impulse response is likely to be poorly estimated. In FIG. 6, a voice synthesizing circuit 23 is provided to prevent this, and the control circuit 19 controls the voice synthesizing circuit 23 during learning of the echo path of the room, for example, when a bell sound is emitted. Is controlled, and an audio signal such as "Denka Katsutsu Taimas" is generated. This audio signal is supplied to the mixing circuit 15 together with the bell sound signal and is emitted from the speaker 2 together with the bell sound. In the case of a dial tone, a voice signal which causes a voice corresponding to the number of the dial tone, for example, a voice "Ichi" in the case of "1" is input to the mixing circuit 15 so as to be superimposed on the dial tone. As a result, the effect that the input error of the telephone number is immediately found is also obtained. At the time of a ringing tone, an audio signal that causes a voice of “Tadai Mayo Bidasichiyudes” is input to the mixing circuit 15 so as to be superimposed on the ringing tone. In this way, by outputting the meaning of the necessary audible sound as an audio signal, it is possible to study the echopath of the room without causing any discomfort to the user, and these sounds are Since it consists of many frequency components, the impulse response of the room can be estimated with sufficiently high accuracy.

In the above description, the audio signal is overlaid on the audible sound, but the audible sound may be stopped and only the audio signal from the audio synthesis circuit 23 may be output.

Although the embodiment of the present invention has been described by taking the in-home loudspeaker telephone as an example, the present invention is not limited to this, and may be used for a car telephone or the like. When used as a car telephone, the subscriber line 7 is a wireless path, and transmission and reception are assigned to independent wireless channels. Therefore, there is no car telephone terminal corresponding to the hybrid circuit 6 in FIG. However, when an automobile telephone and a regular telephone are connected, there is a hybrid circuit in the exchange to which the ordinary telephone is connected, and the speaker loop and the microphone of the automobile telephone and this hybrid circuit can form a signal loop as shown in FIG. It is the same. Also, on the car phone,
Audible sounds such as bell sound and ringing sound exist as necessary for telephone communication.

[Effects of the Invention] As described above, according to the present invention, in the echo canceling circuit, in order to perform the learning of the echo path by using the audible sound generated prior to the call and which the user normally experiences, the call start is started. From time to time, a sufficient echo canceling effect is obtained, and good communication is possible, and no unpleasant sound is generated during the learning.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of a loudspeaker according to the present invention, FIGS. 2 and 3 are flow charts showing its operation, and FIGS. 4, 5, and 6 are loudspeakers according to the present invention. FIG. 7 is a block diagram showing another embodiment of the telephone set, and FIG. 7 is a principal block diagram showing an example of a conventional public address telephone set. 1 ... Microphone, 2 ... Speaker, 3 ... Echo canceling circuit, 8 ... Switch, 9 ... Dial tone generation circuit, 11 ... Call signal detection circuit, 12 ... Bell sound generation circuit, 13 ... Ringing tone detection circuit, 14 ... Busy tone detection circuit, 15 ... Mixing circuit, 16 ... Tap coefficient holding circuit, 20 ... Selection switch, 21 ... Non-linear circuit, 22 ... Selection switch, 23 ... Voice Synthesis circuit.

Claims (4)

[Claims] 1. A transmission signal path including a microphone for transmission, a reception signal path including a speaker for reception, and an echo path generated by coupling the transmission signal path and the reception signal path. An echo canceling circuit for learning and canceling echo signals mixed in the output signal of the microphone;
A loudspeaker telephone equipped with an audible sound signal generating means for supplying an audible sound signal generated for a reception request to the reception signal path and generating an audible sound from the speaker, in a period during which the speaker generates the audible sound. A switch is provided that opens the transmission signal path and closes the transmission signal path during a call period, and the echo canceling circuit is configured to enable learning of the echo path by the audible sound before the start of the call. Loud phone. 2. The echo canceller according to claim 1, further comprising means for holding echopath data obtained by the echo canceller learning with the audible sound, and storing the echopath data held by the means. A loudspeaker telephone characterized by using the initial data of the echo canceling circuit during a call. 3. A loudspeaker telephone according to claim 1 or 2, wherein the audible sound signal is supplied to the reception signal path through a non-linear circuit. 4. The audible sound signal generating means according to claim 1 or 2, further comprising a voice synthesizing means, wherein the audible sound signal for transmitting and receiving requests prior to a call. At the same time, an audible sound signal from the voice synthesizing means is supplied to the reception signal path.