JPH07121027B2 - Loud phone - Google Patents

Description

Description: [Industrial field of application] The present invention relates to a loudspeaker telephone capable of communicating with a speaker and a microphone without using a handset (handset).
In particular, the present invention relates to a loudspeaker telephone having an echo canceling circuit that prevents howling that occurs when a sound emitted from a speaker is reflected by a wall in a room and enters a microphone.

[Conventional technology]

A loudspeaker phone, which allows you to talk using a speaker and a microphone without using a handset (handset), can use both hands freely even during a call, so it is necessary to use it as a car phone, which has been attracting attention in recent years, from the viewpoint of safety. It is also essential to use it as a conference call.

In such a loudspeaker telephone, a reception signal input from a public line through a hybrid circuit (a circuit that performs 2-line to 4-line conversion) is amplified by a reception amplifier and supplied to a speaker, and also output from a microphone. The transmitted speech signal is amplified by a transmission amplifier and then supplied to the public line via the hybrid circuit. At that time, the sound emitted from the speaker is reflected by the wall in the room or the window glass in the car (hereinafter,
This reflected sound is called echo sound. ), When picked up by a microphone, the reverberant sound becomes a sound signal and is amplified by the transmission amplifier, and a part of this sound signal (hereinafter referred to as the reverberation signal) leaks to the reception signal in the hybrid circuit. Then, the echo signal is amplified by the reception amplifier and supplied to the speaker, and as a result, a loop of the echo signal is generated.

Therefore, if a loudspeaker telephone is simply configured with a microphone, a speaker, an amplifier, etc., howling may occur due to the loop of the reverberation signal, making it impossible to communicate.

Conventionally, in order to prevent this howling, a voice switch system has been used in a loudspeaker telephone.

This is because a loss is inserted in the reception signal path (the signal path from the hybrid circuit to the speaker) during transmission (specifically, an attenuator is inserted), and conversely, the transmission signal during reception. This is a method in which a loss is inserted in the path (the signal path from the microphone to the hybrid circuit) to break the loop of the echo signal.

However, in this system, a loss occurs in any one of the signal paths, so that the simultaneous communication becomes impossible. Further, there is a drawback that the beginning and end of the speech is cut off by switching the insertion of the loss, which gives an unnatural feeling to the speech.

As an alternative to this method, the echo cancellation method has been attracting attention in recent years. Note that, as a system related to this echo cancellation system, for example, one described in JP-A-61-123324 can be cited.

This echo cancellation method is a method of canceling only the echo signal obtained from the echo sound in the transmission signal output from the microphone, that is, the transmission path of the sound from the speaker to the microphone (hereinafter referred to as the echo path). .) To create a pseudo echo signal that simulates the echo signal, and subtracts the pseudo echo signal from the transmission signal output from the microphone to eliminate the echo signal.

Accordingly, the echo signal loop described above is broken between the microphone and the hybrid circuit, and howling is prevented. Also, in this system, no loss is inserted into any of the signal paths unlike the voice switch system, so that a simultaneous call can be made, and the beginning and end of the talk are not cut off, and good call quality can be obtained.

However, in this method, for example, when the other party (hereinafter, far-end speaker) and the other party (hereinafter, near-end speaker) side of the telephone talk at the same time (hereinafter, double-talk Two-way call]), the microphone of the loudspeaker on the near-end speaker side has a near-end speaker's voice (that is, reverberant sound) reflected by a wall in a room and emitted from the speaker. Since the voices of the far-end speakers are input at the same time, when estimating the echo path described above, the voices of the near-end speakers input to the microphone are disturbed and accurate estimation cannot be performed. A pseudo echo signal that accurately simulates the echo signal cannot be obtained. As a result, in order to eliminate the echo signal, this inaccurate pseudo-echo signal is reduced from the transmission signal output from the microphone, thereby superimposing an unnecessary signal on the transmission signal and causing noise. There was a problem that it would occur.

Therefore, in order to suppress the noise generation, generally, it is detected whether or not the state is the double-talk state, and when the state is the double-talk state, the operation of estimating the echo path is stopped.

As a method of detecting this double talk, conventionally, the power of the transmission signal output from the microphone and the power of the reception signal input to the speaker are detected, and the two are compared to determine the power of the transmission signal as the reception signal. A certain amount of power that is greater than the power of the reverberant signal (that is, the signal obtained by the reverberant voice) in addition to the reverberant signal (that is, the signal obtained by the voice of the near-end speaker) Existed, that is, it was determined that there was a double talk state. Incidentally, the above-mentioned Japanese Patent Laid-Open No. 61-123324 basically describes the method of detecting this double talk, that is, the statistical characteristics of the call voice (call voice in a conference etc.) Is added.

As described above, the conventional double-talk detection method requires a certain amount of time integration because of the relative comparison between the power of the received signal and the power of the transmitted signal, which causes a delay in the detection time. However, there is a problem that the circuit becomes complicated due to the countermeasure.

Further, since the relative power is compared as described above, the loudness of the voice of the near-end speaker and the loudness of the voice of the far-end speaker (hereinafter,
It is called the vocalization level. ) Is almost the same, it is possible to detect with high sensitivity, but if the pronunciation level of the near-end speaker and the utterance level of the far-end speaker are significantly different, there is a problem of false detection. . For example, when using a loudspeaker phone in a telephone conference where hands-free is essential during voice calls, considering that it is a conference, there is a high probability that multiple people will talk at the same time, and the difference in utterance level between people will vary greatly. To do. Therefore, in such a case, as described above, there is a possibility that the level difference between the utterance level of the near-end speaker and the utterance level of the far-end speaker may become large, and erroneous detection may occur.

By the way, in a telephone conference, etc., in addition to the above-mentioned hands-free being indispensable during a voice call, in addition to voice, the television (TV) image data of a facsimile (FAX), handwritten characters and graphics, and other drawn images. It is necessary to simultaneously transmit non-voice data such as data or personal computer data.

In the conventional conference call system, the voice system and the non-voice system are connected using independent lines. Therefore, the line use cost is high, and the system is expensive and large.

Therefore, as a telephone capable of transmitting drawing image data (a type of non-voice data) at the same time as voice at low cost by using a single line, for example, the one described in Japanese Patent Laid-Open No. 57-75055 or the postal service. There is a telewriting terminal described in the "Survey Report on Telewriting" issued by the Ministry in April 1986. Such a telephone uses a general analog voice telephone line to inexpensively transmit drawn image data at the same time as voice. As this transmission method, a method is used in which a limited audio frequency band is divided, and an audio signal and handwritten image data such as characters and figures are multiplexed and simultaneously transmitted.

Such a telephone is expected to be applied to a simple conference call that requires handwritten information at the same time as voice such as a design meeting. However, in the above-mentioned two proposed examples, the hands-free call means is used. Not disclosed,
Even when handwriting, it was necessary to hold the handset in one hand and make a voice call, which significantly impaired the natural operation of the conference.

[Problems to be Solved by the Invention]

As described above, in the conventional loudspeaker telephone using the echo cancellation method, when double talk is detected, the circuit configuration for detecting the double talk becomes complicated, and in the near-end talk such as the telephone conference. When the level difference between the utterance level of the person speaking and the utterance level of the far-end speaker becomes large, there is a problem that erroneous detection is performed.

Further, in a conventional telephone that can transmit non-voice data such as drawing data at the same time as voice, hands-free communication cannot be performed.

The present invention has been made in view of the above-mentioned problems of the conventional technology, and an object of the present invention is mainly to detect double talk with a simple circuit configuration, regardless of the speaking level of the speaker. It is an object of the present invention to provide a loudspeaker telephone using an echo cancellation method that can be performed more accurately and with higher sensitivity.

[Means for Solving the Problems]

In order to achieve the above-mentioned object, according to the present invention, a speaker and a microphone arranged in the same room and a predetermined frequency band (hereinafter referred to as a rejection band) of a reception signal guided from the public line to the speaker. First filtering means for removing only the signal component, first level detecting means for detecting the level of the reception signal, and a transmission signal output from the microphone are input, and a pseudo echo signal from the transmission signal. Inputting the transmission signal output from the microphone, and subtracting means for reducing the signal to the public line, and only the signal component of the transmission signal in the same frequency band as the removal band of the first filtering means. Second filtering means for passing through, a second level detecting means for detecting the level of the output signal from the second filtering means, and a sound transmission path from the speaker to the microphone. Parameter updating means for sequentially updating a pseudo echo path parameter including a simulated echo path according to the reception signal guided from the public line to the speaker and the output signal from the subtraction means, and the updated pseudo echo path. And a pseudo echo signal creating means for creating the pseudo echo signal on the basis of a parameter.
It is adapted to detect whether or not the state is the double talk state by the output signal from the level detecting means. Then, in the double talk state, the updating of the pseudo echo path parameter in the parameter updating means is stopped.

[Action]

The first filtering unit removes only a signal component in a predetermined frequency band that does not hinder voice communication, from the frequency band of the reception signal guided from the public line to the speaker. Further, the second filtering means inputs the transmission signal output from the microphone, and passes only the signal component of the transmission signal in the same frequency band as the removal band of the first filtering means.

Therefore, the reception signal from which the removal band signal component has been removed by the first filtering means is uttered as voice from the speaker, reflected by a wall in the room, and then collected as echo sound by the microphone. Even if it is output from the microphone as an echo signal, no output is generated in the second filtering means.

However, when the near-end speaker utters a voice, since the signal component of the removal band is not removed from the voice, the voice is collected by the microphone and output as a voice signal from the microphone. Producing an output to the filtering means.

Therefore, by detecting the level of the output signal from the second filtering means by the second level detecting means,
It can detect whether the near-end speaker is making a voice,
Thereby, when the presence of the received signal can be detected by the first level detector when it is confirmed that the near-end speaker is making a voice, it can be reliably determined that the state is the double talk state.

As described above, according to the present invention, the double talk is detected not by the relative comparison between the power of the received signal and the power of the transmitted signal as in the conventional case but by the absolute comparison of the signal levels. Therefore, it can be realized with a simple circuit configuration, and more accurate and more sensitive detection can be performed regardless of the utterance level of the speaker, and erroneous detection does not occur unlike the conventional case.

〔Example〕

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

FIG. 1 is a block diagram showing a loudspeaker telephone system using an echo cancellation system as a first embodiment of the present invention.

In FIG. 1, 1 is a microphone that collects transmitted voice, 2 is a speaker that emits received voice, 3 is a general subscriber line (public line), 4 is a hybrid circuit that performs 2-wire to 4-wire conversion, and 5 is A receiving amplifier that amplifies the receiving signal, 6 is a transmitting amplifier that amplifies the transmitting signal, 7 is a band eliminate filter (hereinafter abbreviated as BEF) that removes only signal components in a predetermined frequency band, and 8 is BEF7. A bandpass filter (hereinafter abbreviated as BPF) that passes only signal components in the same frequency band as the frequency band to be removed, 9 is an echo canceling circuit, 10 and 11 are level detecting circuits, 12 is level detecting circuit 1
A determination circuit that determines whether or not a double-talk state is present, based on output signals from 0 and 11. In addition, the echo canceller 9
Is a reception signal storage circuit 901 that stores a reception signal, and a pseudo echo path parameter storage circuit 90 that stores parameters that simulate the echo path (hereinafter referred to as pseudo echo path parameters).
2, and a superimposition integration circuit 903 that creates a pseudo echo signal that simulates the echo signal from the receive signal stored in the receive signal storage circuit 901 and the pseudo echo path parameter stored in the data of the pseudo echo path parameter storage circuit 902. , The correction amount calculation circuit 904 that calculates the correction amount of the pseudo echo path parameter, and the microphone 1 using the pseudo echo signal generated by the superposition integration circuit 903.
And a subtractor 905 that subtracts the received signal from the.

In the loudspeaker telephone shown in FIG. 1, the reception signal input from the general subscriber line 3 through the hybrid circuit 4 is amplified by the reception amplifier 5 and supplied to the speaker 2.
The transmission amplifier 6 outputs the transmission signal output from the microphone 1.
After being amplified by, it is supplied to the general subscriber line 3 through the hybrid circuit 4. At that time, if the sound emitted from the speaker 4 is reflected by a wall in the room or a window glass in the vehicle and picked up by the microphone 1, the sound (that is, the reverberant sound) becomes a reverberant signal and the transmitting amplifier 6 Is amplified by
When a part of the echo signal leaks into the reception signal in the hybrid circuit 4, the echo signal is amplified by the reception amplifier 5 and supplied to the speaker 2. As a result, a loop of the echo signal occurs. When a loop of the echo signal occurs in this manner, howling occurs due to this loop, and there is a possibility that a call cannot be made.

Therefore, an echo canceling circuit 9 is provided in order to prevent howling caused by such a loop of the echo signal. The operation of the echo canceller 9 will be described below.

First, the operation when not in the double talk state will be described.

In the echo canceling circuit 9, first, the reception signal storage circuit 901
Temporarily stores the received signal x (t) obtained from the general subscriber line 3.

Next, when the received signal x (t) is output as sound from the speaker 2, reflected by a wall in the room, and input to the microphone 1, it becomes a reverberation signal y (t). Therefore,
The echo signal y (t) is expressed as the received signal x (t) multiplied by the transfer function of the echo path (that is, the audio transmission path from the speaker 2 to the microphone 1). As is well known, since the transfer function expressed in the time domain is the impulse response, the echo signal y (t) is the received signal x.
It is a superposition integral of (t) and the impulse response of the echo path.

The pseudo echo path parameter storage circuit 902 temporarily stores the impulse response parameter (that is, the above-mentioned pseudo echo path parameter) _i (t) that approximates the echo path. The impulse response parameter _i (t) that approximates this echo path and the received signal x (t) are superposed and integrated by the superposition integration circuit 903 to create a pseudo echo signal (t). The pseudo echo signal (t) is subtracted from the transmission signal output from the microphone 1 in the subtractor 905, and the echo signal y (t) included in the transmission signal is canceled. still,
In this case, since it is explained that the signal is not in the double talk state, the signal included in the transmission signal is only the echo signal y (t).

On the other hand, the value of the impulse response parameter _i (t) stored in the pseudo echo path parameter storage circuit 902 is initially zero, but the output signal e (t) of the subtractor 905 and the reception signal x (t) are output.
From the above, the correction amount calculation circuit 904 obtains the successive parameter correction amount Δ _i (t), which is stored in the pseudo echo path parameter storage circuit 902 as the impulse response parameter _i (t).
Impulse response parameter _i (t)
Correct the value of. Then, finally, the impulse response parameter _i (t) stored in the pseudo echo path parameter storage circuit 902 approximates the impulse response of the echo path. The correction amount calculation circuit 904 calculates the parameter correction amount Δ _i (t) by calculating the output signal e of the subtractor 905.
The time-square mean value of (t) is used as an evaluation function, and this is performed so as to be the minimum. LMS as this algorithm
Method (Least Mean Square
Method]) or learning identification method is well known.

The operation described above is performed by, for example, (1) to
It is expressed by equation (4).

e (t) = y (t) − (t) (1) Δ _i (t) = α · e (t) · x (t−i) (2) _i (t) = _i (t -1) ＋ Δ _i (t) …… (3) However, t is time and α is a correction coefficient, which is usually set in the range of 0 to 1. N is the number of impulse response parameter data, and i is the number of impulse response parameter data (i = 0, 1, 2, ..., N-1). That is, the impulse response parameter Δ _i (t) consists of N pieces of data,
In the pseudo echo path parameter storage circuit 902, each is stored in N stages of storage units.

On the other hand, in the learning identification method, equation (2) is changed to equation (5).

From equation (1), the pseudo echo signal (t) is the echo signal y.
If (t) is completely approximated (that is, y (t) =
(T)), the output signal e (t) = 0 of the subtractor 905, that is, the input signal to the transmission amplifier 6 becomes zero. That is, in this case, the echo signal y (t) is completely canceled by the pseudo echo signal (t) and is not output from the subtractor 905, so that the above-described echo signal loop does not occur, and therefore howling is caused. There is no.

By the way, the correction of the value of the pseudo echo path parameter (that is, the impulse response parameter approximating the echo path) is not performed in the double talk state (that is, when the sound input to the microphone 1 is only the echo sound), in other words, It needs to be performed when the signal included in the transmission signal output from the microphone 1 is only the echo signal y (t). That is, in the double talk state, the reverberant voice and the voice of the near-end speaker are simultaneously input to the microphone 1,
The echo signal y (t) is masked by the voice signal s (t) of the voice of the near-end speaker, and if the above-mentioned correction is attempted, a large error will occur in the calculation of the parameter correction amount, and eventually the pseudo-echo The value stored in the road parameter storage circuit 902 is significantly changed. As a result, the convergence of the operation of the echo canceling circuit 9 is significantly hindered, and even if the echo canceling circuit 9 once converges, the echo canceling operation is not performed and the howling occurs.

Therefore, in order to prevent this, when the far-end speaker and the near-end speaker speak at the same time, that is, in the double-talk state,
It is necessary to stop the calculation of the parameter correction amount and prevent the value of the pseudo echo path parameter stored in the pseudo echo path parameter storage circuit 902 from being updated.

Next, the following double talk detection direction will be described.

In FIG. 1, the reception signal output from the reception amplifier 5 has a part of its frequency component removed by BEF7. Then, as described above, the reception signal storage circuit 90 of the echo canceling circuit 9
It is supplied to the speaker 1 and is emitted from the speaker 2 into the room. The sound emitted from the speaker 2 is reflected by a wall or the like and is collected by the microphone 1 as echo sound. On the other hand, the BPF8 is a filter that passes only the band component removed by the BEF7.

FIG. 2 (a) shows a schematic band elimination characteristic of the BEF7 shown in FIG. 1, and FIG. 2 (b) shows a schematic band pass characteristic of the BPF8 shown in FIG.

In BEF7, the part to be removed is preferably a part of the audio part that does not hinder the hearing. In Fig. 2 (a), it is 1600Hz.
From the 1900Hz to the 300Hz band.

Now, when in the double talk state, the microphone 1
The sound collected in (1) is the above-mentioned echo sound and the voice of the near-end speaker. Therefore, the transmission signal output from the microphone 1 is composed of the echo signal and the voice signal of the sound of the near-end speaker. Therefore, the echo signal output from the microphone 1 is, as described above, part of the band is removed by the BEF7 when the received signal is the received signal. Therefore, when the transmitted signal is only the echo signal, the BPF8 does not output. . On the other hand, the voice of the near-end speaker is composed of a wide frequency component of the voice and is collected by the microphone 1 without being band-removed. Therefore, the voice signal of the voice of the near-end speaker is included in the transmission signal. When included
BPF8 has output. That is, the output of BPF8 exists only when the near-end speaker utters a voice.

The level detection circuit 10 detects the level of the output signal from the BEF7 to detect the presence of the reception signal, and the level detection circuit
Reference numeral 11 detects the level of the output signal from the BPF 8 to detect the presence of a voice signal by the voice of the near-end speaker. Each output is input to the determination circuit 12, where it is determined whether or not it is in the double talk state. The determination can be easily made by the logical product of the level detection circuits 11 and 12. The determination result of whether or not the state is the double talk state is transmitted to the correction amount calculation circuit 904, and when it is determined that the state is the double talk state, the calculation operation of the parameter correction amount is stopped. As a result, the update of the pseudo echo parameter (that is, the impulse response parameter) stored in the pseudo echo path parameter storage circuit 902 is frozen, and the echo canceling circuit 9 is prevented from malfunctioning.

In FIG. 1, the level detection circuit 10 for detecting the level of the reception signal is connected to the signal path connecting the BEF 7 and the speaker 2. However, the level detection circuit 10 is not limited to this, and may be anywhere in the reception signal path. For example, it may be between the hybrid circuit 4 and the receiving amplifier 5 or between the receiving amplifier 5 and the BEF7.

Further, the BPF 8 may be provided in any transmission signal path, but it is preferable that the BPF 8 is located at the front stage of the echo canceling circuit 9, that is, the position shown in FIG. This is because the echo canceller 9 may generate a frequency component in the band removed by the BEF7, and mistakenly erroneously recognize the erroneous component as a voice signal of the voice of the near-end speaker.
Therefore, for example, between the subtractor 905 and the transmission amplifier 6, BPF8
It is not preferable to connect.

By the way, the performance of the echo canceling circuit 9 depends on the echo canceling amount ERLE.
It is evaluated by (Echo Return Loss Enhancement). this
The ERLE is usually represented by the ratio of the power of the reception signal to the power of the echo signal included in the transmission signal after the echo signal is erased. That is, ERLE is the output power of the reception amplifier 5 and the transmission amplifier 6 when the double-talk state is not set (that is, when the near-end speaker is not making a sound and only the far-end speaker is making a sound). It is the ratio of the input power to the received signal, and indicates how much the echo component of the reception signal (that is, the echo signal) is attenuated before being transmitted.

This ERLE becomes an infinite value if the value of the pseudo echo path parameter matches the value of the actual echo response impulse response parameter and the processing of the echo canceling circuit 9 is ideal.
In other words, the reverberant component of the received signal is completely canceled and is not transmitted. However, in practice, the echo canceling circuit 9 is a finite number of bits of arithmetic processing, or the number of pseudo echo path parameter data is smaller than the actual echo path impulse response parameter data. It is a finite value.

Further, in FIG. 1, the echo canceling circuit 9 creates a pseudo echo signal by using the signal from which a partial band component has been removed by BEF7, and cancels the echo signal. However, originally, in order to create a pseudo echo signal (estimate the echo characteristic in a room), signal components in all bands are necessary. Therefore, in this example, the estimation is incomplete and the ERLE decreases. This decrease in ERLE is BE
The wider the rejection band at F7, the larger. Therefore, it is desirable that the removal band in BEF7 is narrow. However, if it is unnecessarily narrowed, it causes a malfunction in the detection of the voice signal by the voice of the near-end speaker in BPF8. This is because the voice has a harmonic structure based on the fundamental frequency (pitch frequency of 60 to 150 Hz), so it may not be detected in a narrow band.

FIG. 3 is a characteristic diagram showing an example of the output signal spectrum of the BEF7 and the subtractor 905 in FIG.

In FIG. 3, i is the output signal spectrum of BEF7, e is the output signal spectrum of the subtractor 905, and f is the output signal spectrum of the subtractor 905 without VEF7. Note that i and e
Or the level difference from f is ERLE.

As is clear from FIG. 3, when the band is removed by BEF7,
ERLE is worse than ERLE without removal. Further, when the band is removed by BEF7, of the output signal of the subtractor 905 (that is, the echo signal that could not be canceled by the echo canceling circuit 9), the signal component with a high relative level is concentrated near the removal band of BEF7. You can see that

FIG. 4 is a block diagram showing a second embodiment of the present invention.

4, the same reference numerals as those in FIG. 1 indicate the same parts.

In this embodiment, the BEF 7 is connected between the echo canceling circuit 9 and the speaker 2 as shown in FIG.

Since the echo canceling operation and the double talk detecting operation in this embodiment are the same as those in the embodiment of FIG. 1, the description thereof will be omitted.

In the explanation of the operation of the echo canceling circuit 9 described in the embodiment of FIG. 1, the impression that the echo path is only the echo path of the space from the speaker 2 to the microphone 1 is given. Must be considered including the electric circuit. That is, the echo canceling circuit 9 uses the characteristics of the speaker 2 → room space → the signal path of the microphone 1 in the embodiment of FIG. 1 as the echo path, and BEF7 → speaker 2 → room of the embodiment in FIG. It is necessary to estimate the characteristics of the signal path of the space → microphone 1 to cancel the echo signal. That is, the microphone 1 and the speaker 2 (in the embodiment of FIG. 4, BE
It is necessary to estimate the transfer characteristics that combine the transfer characteristics of F7) and the transfer characteristics of the room, create a pseudo echo signal, and eliminate the echo signal. The transfer characteristic is equivalent to the impulse response, and the impulse response for each transfer characteristic will be described below.

Generally, the impulse response of the speaker 2 is 8 ms because of its large diaphragm area and mass, as shown in FIG. On the other hand, since the microphone 1 has a small diaphragm area and a small mass, its impulse response is negligibly shorter than that of the speaker.

Next, as a transmission characteristic of space, FIG. 6 shows an impulse response of the space inside the vehicle, for example. In the figure, the initial flat delay is the propagation delay of the direct sound from the speaker to the microphone. As shown in FIG. 6, the length of the impulse response in space is 28 ms.

As described above, normally, if the impulse response in space is the longest and the impulse response corresponding to this length is considered,
The impulse response of the microphone and speaker is considered to be included in it. Therefore, in the embodiment shown in FIG. 1, only the impulse response of the space need be considered.

Next, we consider the transmission of BEF7 in terms of impulse response in the time domain. According to the theory of digital filters, a filter having a certain transfer characteristic can be composed of a non-recursive digital filter having tap coefficients as impulse response sequences when the characteristic is inverse Fourier transformed. For example, the impulse response of BEF7 having the band elimination characteristic shown in FIG. 7 (a) is as shown in FIG. 7 (b), and the impulse response length in this case is 48.
is ms. Therefore, in the embodiment of FIG. 4, it is necessary to consider the impulse response of BEF7.

Now, assuming that the echo canceling circuit 9 performs the processing operation using the digital data sampled at 8 kHz, the above equation (4) (that is, the superimposition integrating circuit 903 using the reception signal and the pseudo echo path parameter) is used. Pseudo echo signal (t) by superposition integration
As for N in the formula), 28 ms / 125 μs = 224 for the impulse response length of 28 ms in the space shown in FIG.
However, 48 ms / 125 μs = 384 is required for the impulse response length of 48 ms of BEF7 having the band elimination characteristic of FIG. 7 (a).

Therefore, when N = 256 is used as N, the echo canceller 9 can cope with the impulse response of the space, but 32 ms
It cannot support the impulse response of BEF7 with the above length. Therefore, although the echo signal can be almost canceled in the embodiment of FIG. 1, the echo signal cannot be canceled completely in the embodiment of FIG. 4, and the ERLE is deteriorated.

FIG. 8 is a characteristic diagram showing an example of the output signal spectrum of the BEF7 and the subtractor 905 of FIG.

In FIG. 8, i is the output signal spectrum of BEF7, g is the subtracter when N is smaller than the number of tap coefficients of BEF7, and h is the subtracter when N is sufficiently larger than the number of tap coefficients of BEF7. It is an output signal spectrum of 905. Note that i and g
Or the level difference from h is ERLE.

As is clear from FIG. 8, when N is smaller than the number of tap coefficients of BEF7, ERLE is significantly deteriorated.
Therefore, N, that is, the reception signal storage circuit used in the equation (4)
The number of data in each of 901 and the pseudo echo path parameter storage circuit 902 must be at least the number of tap coefficients that approximate the characteristics of BEF7.

If N is sufficiently larger than the number of tap coefficients of BEF7, the signal component having a high relative level in the output signal of the subtractor 905 (that is, the echo signal that could not be canceled by the echo canceling circuit 9) is removed by BEF7. It can be seen that they are concentrated near the band.

FIG. 9 is a block diagram showing a third embodiment of the present invention.

9, the same reference numerals as those in FIG. 1 indicate the same parts. In addition, 13 is BEF and its band elimination characteristic is the 10th.
As shown in FIG. 7A, the width of the removal band is wider than the width of the removal band of BEF7 (300 Hz band) shown in FIG.

In this embodiment, the BEF 13 removes, from the output signal of the subtractor 905, a signal component having a high relative level, which is concentrated in the vicinity of the removal band of BEF 7 shown by e in FIG. Prevents the far-end speaker from listening. It also prevents the decrease of ERLE near the BEF7 removal band. Other operations are similar to those of the embodiment shown in FIG.

It is obvious that the BEF 13 may be composed of two BEFs having band elimination characteristics centering on the corner frequencies (1600Hz, 1900Hz) of the BEF 7, as shown in FIG. 10 (b). FIG. 11 is a block diagram showing the fourth embodiment of the present invention.

In FIG. 11, the same symbols as those in FIG. 9 indicate the same items.

In this embodiment, BEF13 shown in FIG. 9 is added to the embodiment shown in FIG. The operation of this BEF13 is the same as that of the embodiment of FIG. 9 among the output signals of the subtractor 905, h of FIG.
The signal component with a high relative level, which is concentrated in the vicinity of the BEF7 removal band shown in, is removed, and this signal component is prevented from being heard by the far-end speaker as noise. Other operations are first
This is similar to the illustrated embodiment.

FIG. 12 is a block diagram showing the fifth embodiment of the present invention.

In FIG. 12, the same symbols as in FIG. 9 indicate the same items. In addition, 14 is a mixer, 15 and 16 are BPFs, 17 is a modulation / demodulation circuit, and 18 is a data signal line.

In this embodiment, non-voice data is inserted in the BEF7 removal band to perform data communication. That is, the data signal line
The non-voice data obtained from 18 is modulated by the modulation / demodulation circuit 17, filtered by the BPF 15, mixed by the output signal of the BEF 13 by the mixer 14, and transmitted from the hybrid circuit 4 to the general subscriber line 3 via the transmission amplifier 6. Is sent to the far-end speaker. On the contrary, the non-voice data sent from the far-end speaker is the general subscriber line 3
Via hybrid circuit 4 and receiver amplifier 5 to BPF16
The data signal line 18
Be led to.

In the present embodiment, the frequency of the modulated signal obtained by modulating the non-voice data is limited to the BEF7 removal band and transmitted. for that reason,
The modulated signal of the non-voice data transmitted from the far-end speaker is removed by BEF7 after being output from the reception amplifier 5, and is not heard by the speaker 2. In addition, the BEF13 does not mix the voice signal of the voice of the near-end speaker with the modulated signal transmitted to the far-end speaker.

In addition, echo canceling operation and double talk detection operation
Since it is the same as the embodiment shown in FIG. 9, its explanation is omitted.

As described above, according to the present embodiment, it is possible to simultaneously perform the hands-free call and the data communication by the stable echo canceling operation.

FIG. 13 is a block diagram showing a sixth embodiment of the present invention.

In FIG. 13, the same symbols as those in FIG. 12 indicate the same items. In addition, 19 is modulated non-voice data (modulated signal)
Is a carrier detection circuit for detecting the presence of the carrier, and 20 is an OR circuit for taking the logical sum.

Similar to the embodiment shown in FIG. 12, the present embodiment effectively uses the removal band of BEF7 and enables data communication at the same time as hands-free communication.

Since the BEF 7 is not connected to the echo canceling circuit 9 before the echo canceling circuit 9, a modulated signal of non-voice data is input. When the modulation signal is input to the echo canceling circuit 9, the modulation signal transmitted from the far-end speaker is input in addition to the case where the modulation signal transmitted from the far-end speaker is input by the hybrid circuit 4. The received signal may be leaked and input to the roadside. It is known that the echo canceller 9 malfunctions when a narrow band signal such as this modulated signal is input.

Therefore, in this embodiment, the carrier detection circuit detects the presence of the modulation signal transmitted from the far-end speaker or transmitted to the far-end speaker.
If it is detected in step 19 and exists, the operation of the correction amount calculation circuit 904 is stopped. Therefore, the correction amount calculation circuit 904 calculates the logical sum of the output signal of the determination circuit 12 that detects double talk and the output signal of the carrier detection circuit 19 that detects the presence of the modulated signal in the OR circuit 20, and outputs The signal will control the operation.

As described above, according to this embodiment, similarly to the embodiment of FIG. 12, it is possible to simultaneously perform the hands-free call and the data communication by the stable echo canceling operation.

Now, when an external noise signal n (t) having no correlation with the received signal (for example, vehicle noise, external sound such as radio) is mixed in the echo path from the speaker 2 to the microphone 1, a steady state is generated. ERLE is (Noda: “Effects of noise signal and parameter variation on learning identification method” measurement and control vol.8, No.5, P303-P31
2, see May 1969).

When the external noise signal n (t) is present in the echo path and its amount becomes large, ERLE decreases as shown in the equation (6), the echo canceling operation becomes insufficient, and howling occurs.

FIG. 14 is a block diagram showing a seventh embodiment of the present invention.

14, the same reference numerals as those in FIG. 1 denote the same parts. In addition, 21 is an echo noise ratio calculation circuit, 22 is an attenuator control circuit, and 23 and 24 are attenuators.

The present embodiment is a combination of the echo canceling circuit 9 and the echo blocking circuit used in the above-mentioned voice switch system. The echo blocking circuit includes an echo noise ratio detection circuit 21, an attenuator control circuit 22, and attenuators 23 and 24.

Since the running noise and the traffic noise are input to the microphone 1 inside the vehicle, the ERLE of the echo canceling circuit 9 is lowered as shown in the equation (6). Therefore, assuming that the ERLE required to prevent howling is 60 dB, if the ERLE of the echo canceling circuit 9 becomes 50 dB due to the mixing of noise, it is 10 dB.
There will be a shortage of dB. Attenuators 23 and 2
The feature of the present embodiment is to give in 4.

Now, the operation of this embodiment will be described.

When the near-end speaker is not making a sound, the echo noise ratio calculation circuit 21 outputs the echo signal y (t) by the output signal of the level detection circuit 10 and the output signal of the level detection circuit 11.
And the power ratio of the external noise signal n (t) (hereinafter referred to as the echo noise ratio) can be roughly obtained.

That is, in the past, it was difficult to separate the echo signal and the external noise signal, but in the present embodiment, since the BEF7 removes a part of the band of the received signal, the speaker 2 outputs the sound. Even if the echo sound and noise are collected by the microphone 1, if the near-end speaker does not make a sound, the BPF8
Since only the external noise signal is output to the output of, the echo signal and the external noise signal can be reliably separated, and therefore, the above-mentioned echo noise ratio can be obtained.

The echo noise ratio calculating circuit 21 estimates the ERLE of the echo canceling circuit 9 from the obtained echo noise ratio based on the equation (6), notifies the attenuator control circuit 22 of the insufficient ERLE, and at the same time, in the transmitting state. Information indicating whether there is an incoming call or a receiving state is also notified.

The attenuator control circuit 22 controls the attenuators 23 and 24 based on these deficient ERLEs and information indicating whether it is in the transmitting state or the receiving state. That is, in the case of the receiving state, the attenuator 24 gives the transmission signal an attenuation equal to the insufficient ERLE, and the attenuation amount of the attenuator 23 is zero. Also, in the case of the transmitting state, the attenuator 23 has a shortage of ER.
Attenuator is designed to give the received signal equal attenuation to LE.
The attenuation of 24 is set to zero.

According to the present embodiment, since the lack of ERLE of the echo canceling circuit 9 is compensated for even in a noisy environment such as in a vehicle, stable hands-free communication can be performed.

FIG. 15 is a block diagram showing the eighth embodiment of the present invention.

In FIG. 15, the same symbols as those in FIG. 14 indicate the same items. In addition, reference numeral 25 is a voice detection circuit for detecting whether or not the voice of the near-end speaker is collected by the microphone 1.

This embodiment is a combination of the echo canceling circuit 9 and the echo canceling circuit used in the voice switch system as in the embodiment of FIG. The present embodiment is different from the embodiment shown in FIG. 14 in that a voice detection circuit 25 is newly provided and the connection point between the level detection circuit 10 and the attenuator 23 is changed.

As shown in FIG. 15, if the level detection circuit 10 and the attenuator 23 are in the reception signal path, the same effect as the embodiment of FIG. 14 can be obtained regardless of the connection position.

Further, the voice detection circuit 25 detects a voice signal among signals other than the echo signal y (t) output from the microphone 1 (that is, the voice signal of the voice of the near-end speaker and the external noise signal of noise). It is a thing. As a detection principle, a well-known correlation method is used in voice processing.

Therefore, although the voice of the near-end speaker may be mistaken as noise in the embodiment of FIG. 14, according to the present embodiment, the voice detection circuit
It is possible to accurately estimate the reverberant noise ratio by distinguishing between the voice and noise of the near-end speaker by 25.

〔The invention's effect〕

According to the present invention, since the double talk is detected by the absolute comparison of the signal levels, not by the conventional relative comparison between the power of the received signal and the power of the transmitted signal, a simple circuit is provided. In addition to being realized by the configuration, the detection can be performed more accurately and with high sensitivity regardless of the voice level of the speaker, and erroneous detection does not occur unlike the conventional case. Therefore, it is possible to provide a loudspeaker telephone suitable for a hands-free call in a telephone conference (telephone conference such as video conference).

Further, when the non-voice data is inserted and transmitted within the removal band of the filtering means for removing the predetermined band of the received signal, stable hands-free call and data communication can be performed at the same time.

Further, when the voice switch system is combined, stable hands-free call can be performed even under high noise.

[Brief description of drawings]

FIG. 1 is a block diagram showing the first embodiment of the present invention, and FIG.
Figure (a) is a characteristic diagram showing the band elimination characteristics of BEF7 in Figure 1.
2 (b) is a characteristic diagram showing the band pass characteristic of the BPF 8 in FIG. 1, FIG. 3 is a characteristic diagram showing an example of the output signal spectrum of the BEF7 and the subtractor 905 in FIG. 1, and FIG. FIG. 5 is a block diagram showing a second embodiment of the invention, FIG. 5 is an explanatory diagram showing an example of an impulse response of the speaker 2 of FIGS. 1 and 4, and FIG. 6 is an example of an impulse response of a space in a vehicle. Explanatory drawing, FIG. 7 (a) is a characteristic diagram showing the band elimination characteristic of BEF7 of FIG. 4, and FIG. 7 (b) is an example of the impulse response of BEF having the band elimination characteristic of FIG. 7 (a). Explanatory drawing which shows, 8th
4 is a characteristic diagram showing an example of the output signal spectrum of the BEF7 and subtractor 905 in FIG. 4, FIG. 9 is a block diagram showing the third embodiment of the present invention, and FIGS. 10 (a) and 10 (b) are Figure 9 of each
FIG. 11 is a characteristic diagram showing the band elimination characteristic of BEF13, FIG. 11 is a block diagram showing the fourth embodiment of the present invention, and FIG. 12 is the fifth embodiment of the present invention.
FIG. 13 is a block diagram showing a sixth embodiment of the present invention, FIG. 14 is a block diagram showing a seventh embodiment of the present invention, and FIG. 15 is a block diagram showing the present invention. It is a block diagram which shows the Example of 8. Explanation of reference numerals 1 ... Microphone, 2 ... Speaker, 7 ... BEF, 8 ... BPF,
9 ... Echo canceling circuit, 10, 11 ... Level detecting circuit, 12 ... Judgment circuit, 901 ... Received signal storage circuit, 902 ... Pseudo echo path parameter storage circuit, 903 ... Superposition integration circuit, 904 ... Correction amount calculation circuit.

Claims (5)

[Claims] 1. A speaker and a microphone arranged in the same room, and a first component for removing only a signal component of a predetermined frequency band (hereinafter referred to as a removal band) of a received signal guided from a public line to the speaker. Filtering means, first level detecting means for detecting the level of the reception signal, and a transmission signal output from the microphone, subtracts a pseudo echo signal from the transmission signal, and guides the signal to the public line. A second means for inputting the transmission signal output from the subtraction means and the microphone, and for passing only the signal component of the transmission signal in the same frequency band as the removal band of the first filtering means.
Filtering means, second level detecting means for detecting the level of an output signal from the second filtering means, and a pseudo echo path for simulating an echo path including a sound transmission path from the speaker to the microphone. Parameter updating means for sequentially updating parameters in accordance with the received signal guided to the speaker from the public line and the output signal from the subtracting means,
Pseudo echo signal creating means for creating the pseudo echo signal on the basis of the updated pseudo echo path parameter, and whether the two-way communication state is based on the output signals from the first and second level detecting means. A loudspeaker telephone characterized in that it is detected whether or not the update of the pseudo echo path parameter in the parameter updating means is stopped when in a two-way communication state. 2. The loudspeaker telephone set according to claim 1, wherein the signal guided from the subtracting means to the public line has a frequency band including a frequency band including the elimination band of the first filtering means and wider than the elimination band. A loudspeaker telephone having a third filtering means for removing a signal component. 3. The loudspeaker telephone according to claim 2, wherein the modulating means for modulating non-voice data and outputting the modulated signal, and the modulated signal output from the modulating means are output from the third filtering means. Mixing means for mixing with a signal, and demodulation means for demodulating the modulated signal mixed with the reception signal guided from the public line to the first filtering means to obtain non-voice data, A loudspeaker telephone characterized in that the frequency band is within the removal band of the first filtering means. 4. The loudspeaker telephone according to claim 3, further comprising carrier detection means for detecting a modulation signal output from said modulation means and a modulation signal mixed with said reception signal, and detection by said carrier detection means. As a result, when the modulation signal is output from the modulation means or when the reception signal is mixed with the modulation signal, the updating of the pseudo echo path parameter in the parameter updating means is stopped. Loud phone. 5. The loudspeaker telephone according to claim 1, wherein the power of the echo signal included in the transmission signal output from the microphone is based on the output signals from the first and second level detecting means. And an external noise signal power ratio (hereinafter referred to as "echo noise ratio"), and a first attenuator for attenuating a reception signal guided from the public line to the speaker. A second attenuator for attenuating the signal introduced from the subtracting means to the public line, and controlling the attenuation amounts of the first and second attenuators according to the calculated reverberation noise ratio. A loudspeaker telephone characterized by the above.