JP4735419B2 - Voice communication device - Google Patents

Description

  The present invention is used in a voice call device that maintains a call state according to noise.

  The voice communication device is a device for two parties in different places (referred to as the near end and the far end) to talk to each other, and one of the two parties is the near end speaker. The person at the far end is called the far end speaker. Also, the end where the far-end speaker speaks and the far-end audio signal recorded by the microphone at the far end is input is called the receiving input end, and the audio signal input to the receiving input end is the speaker at the near end. The end that is output by is called the receiving output end, and the end where the near-end speaker utters and receives the near-end audio signal recorded by the microphone at the near-end is called the transmission input end. The end where the audio signal input to the input end is output by the far-end speaker is called a transmission output end.

  In the present invention, the audio signal is referred to as an audio signal including background noise as well as a voice emitted by a person. In addition, a case where a person does not utter a voice but only background noise is also called a voice signal.

  In such a voice communication device, for example, the far-end speaker's voice output from the near-end speaker is input to the near-end microphone, reaches the transmission output end at the far-end, and is output from the far-end speaker. In some cases, this is repeatedly input to the far-end microphone and output from the near-end speaker. When such a situation occurs, the voice of the far-end speaker continues for a while as an echo, and as a result, the voice input of the far-end speaker may be hindered.

  Further, the voice communication device is provided that the gain from the transmission output end to the reception input end is less than 0 dB in a call with the far end, but the gain from the transmission output end to the reception input end is 0 dB ( Alternatively, in a state where the transmission output terminal to the reception input terminal are short-circuited), howling may occur due to an oscillation state caused by a near-end audio signal. As a result, the call may be hindered.

  The same can be said even if the positions of the near end and the far end in the above are interchanged.

  In order to avoid such a situation, the presence or absence of near-end speech and far-end speech is detected, and loss is inserted into the signal path on the near-end or far-end where no speech is detected. To avoid echo or howling. In addition, when voice is detected at both the near end and the far end, echo or howling is avoided by performing loss insertion on either one.

  Furthermore, the above-described loss insertion amount can be reduced by providing an echo canceller that detects that the same audio signal as the audio signal output from the reception output terminal is input from the transmission input terminal and cancels this. it can. The smaller the loss insertion amount, the higher the transmission efficiency of the audio signal and the easier it is to secure the volume required for the call. Therefore, it is desirable that the loss insertion amount is smaller.

Japanese Patent Laid-Open No. 10-243082 JP 2005-347875 A

  The echo canceller in the above-described prior art can be expected to have a sufficient echo cancellation amount when the near-end background noise is relatively small, but should be canceled when the near-end background noise is relatively large. Since it becomes difficult to clearly distinguish the echo and the background noise, the echo cancellation performance is deteriorated and the call is hindered. Furthermore, if the means for detecting the presence or absence of speech erroneously detects background noise as a near-end speaker's transmission, it will cause loss insertion in the signal path between the receiving input end and the receiving output end, It may be difficult to hear the voice of the far-end speaker on the near-end side.

  In order to solve such a problem, it is limited to a solution using an adaptive filter such as a method of adding the input signal on the near end side to the receiving side of the echo canceller (see, for example, Patent Document 1) or a bathroom. For example, a method of using the loss estimation amount and the fixed loss amount properly depending on the near-end background noise (see, for example, Patent Document 2) has been proposed.

  However, for example, in the proposal of Patent Document 1, the applicable background noise needs to be clearly distinguishable by a voice signal to be transmitted and a filter, and cannot be applied to various actually existing background noises. Moreover, the proposal of patent document 2 is a proposal specialized for the use in a bathroom, and it is difficult to extend an application range other than a bathroom.

  The present invention has been made under such a background, and an object of the present invention is to provide a voice call device capable of making a good call under various noise environments.

  The present invention provides a transmission signal path and a reception signal path between a transmission input terminal and a transmission output terminal and between a reception input terminal and a reception output terminal, respectively. Voice communication means for making a call between the voice signal detection means for detecting the presence or absence of a voice signal in the transmission signal path and the reception signal path, and detection of the voice signal detection means Loss insertion means for adaptively setting the loss insertion location and loss insertion amount of the transmission signal path or the reception signal path according to the result, provided between the reception output terminal and the transmission input terminal, The required loss insertion amount is estimated based on the echo canceller that cancels the echo generated by the far-end speaker's speech signal, the speech signal after insertion loss insertion and the speech signal after echo cancellation by the echo canceller. A voice communication device that includes a loss amount estimating means for.

  Here, a feature of the present invention is that an average signal strength in a predetermined time of a voice signal input from the transmission input terminal or a voice signal after echo cancellation by the echo canceller is an estimated noise level, According to the present invention, there is provided a state classification means for classifying the quality of the call environment in a plurality of predetermined stages according to the strength of the estimated noise level. As described above, the present invention is characterized in that the quality of the call environment is classified and optimal control is performed under various noise environments.

  For example, the state classification means is low noise according to the strength of the estimated noise level, and an estimated value of loss insertion amount by the loss amount estimation means (hereinafter referred to as an estimated loss amount) is within the applicable range and The voice signal detection means detects the presence / absence of a voice signal and the echo canceller can cancel the echo. The voice signal is low and the estimated loss amount is out of the applicable range and the voice signal is detected by the voice signal detection means. Detection of the presence or absence of speech and the state of speech under noise in which echo cancellation by the echo canceller is possible, detection of the presence or absence of a speech signal by the speech signal detection means, and the estimated loss amount being out of the applicable range when the noise is high There is provided means for classifying into three states of a call state under high noise in which echo cancellation by the echo canceller is impossible.

  By performing the classification in this manner, for example, the loss insertion unit is classified into a low-noise call state by the state classification unit, detects a voice signal in the reception signal path, and detects in the transmission signal path. When no voice signal is detected, a predetermined amount of received power limit loss is inserted into the received signal path, the estimated amount of loss is inserted into the transmitted signal path, and a voice signal is transmitted through the transmitted signal path. And when no voice signal is detected in the received signal path, or when a voice signal is detected in both the received signal path and the transmitted signal path, or in the received signal path and the transmitted signal path. When neither of the voice signals is detected, the greater of the received output limit loss amount or the estimated loss amount is inserted into the received signal path as a loss.

  In other words, the low-noise call state means that the near-end background noise level is low, the transmitted voice is easy to detect, and echo cancellation by the echo canceller is possible (specifically, the adaptive filter in the echo canceller converges) ) And the estimated loss is within the applicable range. When a voice signal is detected in the transmission signal path, no loss is inserted in the transmission signal path, so that the voice signal of the near-end speaker reaches the far end satisfactorily. Also, when voice signals are detected simultaneously on both signal paths, an estimated loss amount larger than the received output limit loss amount may be inserted into the received signal path, so that the far-end speech is somewhat at the near-end. Although it may be small, it is sufficiently audible because the background noise level at the near end is low. The low-noise call state is the best call state with low background noise.

  Further, for example, when the loss insertion means is classified into a call state under noise by the state classification means, detects a voice signal in the transmission signal path, and does not detect a voice signal in the reception signal path, the reception signal Insert the loss of the received output limit loss amount or the first fixed loss amount capable of suppressing a predetermined echo and howling into the path, whichever is larger, in both the received signal path and the transmitted signal path When an audio signal is detected, the received power limit loss amount is inserted into the received signal path as a loss, and the first fixed loss amount is inserted as a loss into the transmitted signal path.

  In other words, in the noisy call state, the probability that the voice signal detection means misdetermines the near-end background noise as the transmitted voice is high, and echo cancellation by the echo canceller is possible, but the estimated loss amount is applied. A state that is out of range. When the background noise level increases, when a voice signal is detected simultaneously on both signal paths, a large loss that is greater than the received output limit loss amount between the receiving input terminal and the receiving output terminal, as in a low noise call state. If inserted, the received output may be erased by background noise and may not be heard. Therefore, in the state of speech under noise, when a far-end voice is input, loss is inserted into the transmission signal path without performing loss insertion exceeding the reception output limit loss amount in the reception signal path.

  At this time, since the estimated loss amount is out of the applicable range, the estimated loss amount cannot be applied as it is. Therefore, the first fixed loss amount is inserted as the minimum necessary loss amount in order to prevent echo or howling.

  As a result, in a noisy call state, when the near-end talker and the far-end talker speak simultaneously, it is difficult for the near-end talker to reach the far end by inserting the first fixed loss amount. Although a situation may also occur, if either the near-end speaker or the far-end speaker talks alternately, it becomes possible to talk without any problem.

  However, there will be measures to enable a call even when the far-end speaker and the near-end speaker cannot talk alternately because the receiving state continues due to a large background noise at the far-end side. It will be necessary. Therefore, in the present invention, a call state under high noise is provided that is applied when the background noise level is higher than that in the call state under noise.

  For example, the loss insertion means is classified into a high-noise call state by the state classification means, detects a voice signal in the transmission signal path, and does not detect a voice signal in the reception signal path, or When a voice signal is detected in both the signal path and the transmission signal path, the reception output limit loss amount is inserted into the reception signal path as a loss, and is smaller than the first fixed loss amount in the transmission signal path. Insert a second fixed loss amount.

  In other words, the speech state under high noise means that the voice signal detection means misjudged the near-end background noise as the transmitted voice, and normal echo cancellation by the echo canceller is impossible, and the loss estimation amount is also large. A condition that falls outside the scope of application.

  When a voice signal is detected simultaneously on both the transmission signal path and the reception signal path in the above-described noise state, the near end is inserted by inserting the first fixed loss amount into the transmission signal path. When a situation occurs where the speaker's voice cannot reach the far-end speaker, the near-end speaker recognizes that it is difficult to communicate with the far-end speaker, Increase the transmission level by bringing the mouth close to the microphone. However, if this still makes it difficult to communicate with the far-end speaker, the near-end speaker takes actions such as making the voice louder or bringing the mouth closer to the microphone. If the active action continues, the estimated noise level increases.

  In such a high noise call state where the estimated noise level is extremely high due to excessive background noise at the near end and the behavior of the near end speaker, the transmission signal path is replaced with the first fixed loss amount. The second fixed loss amount, which is the minimum loss amount that does not perform howling at the near end, is inserted, and the near end speech is transmitted to the far end without being suppressed more than necessary. Therefore, the second fixed loss amount is smaller than the first fixed loss amount.

  As a result, the voice of the near-end speaker can easily reach the far-end speaker, and a call between the near-end speaker and the far-end speaker becomes possible.

  As described above, by performing appropriate loss insertion under various noise environments, a good call can be performed under various environments.

  Further, the present invention can be viewed from the viewpoint of a program. That is, the present invention is a program that, when installed in a general-purpose information processing apparatus, causes the general-purpose information processing apparatus to realize a function corresponding to the voice call device of the present invention. By recording the program of the present invention on a recording medium, the general-purpose information processing apparatus can install the program of the present invention using this recording medium. Alternatively, the program of the present invention can be directly installed on the general-purpose information processing apparatus via a network from a server that holds the program of the present invention.

  Thereby, the voice call device of the present invention can be realized using a general-purpose information processing device.

  ADVANTAGE OF THE INVENTION According to this invention, the voice call apparatus which can perform a favorable call in various noise environments is realizable.

  Embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of the present invention. The voice communication device according to the embodiment of the present invention includes a reception input terminal 1 that receives a far-end speaker signal, a reception output terminal 6 that outputs a far-end voice to the near end, and a transmission that inputs a sound at the near end. An input terminal 5, a transmission output terminal 3 for outputting a near-end speaker voice at the far end, an echo canceller 14 for canceling a near-end echo generated at the near end, and a background noise estimation means for estimating a noise state generated at the near end 21, voice switch unit 10 for controlling the amount of loss to be inserted into the reception signal path or transmission signal path depending on the presence or absence of far-end speaker voice or near-end speaker voice, and reception loss insertion means 11 for inserting loss in the reception signal path Transmission loss insertion means 13 for inserting loss in the transmission signal path, loss amount estimation means 20 for estimating the amount of loss necessary for suppressing howling or echo at the near end, and suppressing the output of the near end speaker Required loss Comprising a receiver output limit estimating means 60 for estimating the certain reception output limit loss 61.

  The echo canceller 14 learns an adaptive filter using FIR type echo simulation filter coefficients that simulate near-end echoes, using the transmission input signal 7 input at the transmission input terminal 5 and the voice signal 12 after insertion of the reception loss as reference signals. And the result of the convolution operation of the reference signal and the echo simulation filter is taken as a pseudo echo.

  Then, the difference between the pseudo-echo from the transmission input signal 7 is used as an error signal, and the result obtained by dividing the power of the absolute value component of this error signal by the absolute value component of the transmission input signal 7 is used as an echo cancellation amount. The monitoring of the echo cancellation amount is started from the start, and it is determined that the target echo cancellation amount, which is an echo cancellation amount that is desired to be reached at the time of convergence, is the non-convergent state until the echo cancellation amount is satisfied, and the subsequent state is the convergence state. 16 as a function of outputting.

  Further, in the non-converged state, the transmission input signal 7 is output as an audio signal 15 after echo cancellation, and in the converged state, the error signal is output as an audio signal 15 after echo cancellation.

  The loss amount estimation means 20 uses the time average power (for example, 1 to 3 seconds) of the absolute value component of the audio signal 12 after the reception loss is inserted as the estimated loss amount reception power, and the absolute value component of the audio signal 15 after echo cancellation. Is the estimated loss amount transmission power.

  Furthermore, a loss margin amount that is a fixed loss amount for sufficiently suppressing echoes and compensating for fluctuations in the near-end echo, a minimum estimated loss amount that is a minimum value of the estimated loss amount 23, and a maximum value of the estimated loss amount 23 The maximum estimated loss amount is determined.

  Then, the initial value of the estimated loss amount reception power is a positive number excluding 0, and the initial value of the loss estimation amount transmission power is a value obtained by multiplying the loss amount estimation reception power by the reciprocal of the second fixed loss amount. When the estimated loss amount reception power is divided by the estimated loss amount transmission power multiplied by the loss margin amount, the result of limiting the range between the minimum estimated loss amount and the maximum estimated loss amount is estimated loss amount 23. To do.

  The estimation procedure of the background noise estimation means 21 is shown in the flowchart of FIG. As shown in FIG. 4, the background noise estimation means 21 is obtained by calculation using the long-term average power of the absolute value component of the audio signal 15 after echo cancellation as the estimated noise level PN (S1), and PNth0 is PNth1. In two threshold values of PNth0 and PNth1 defined by smaller values, when PN is less than PNth0, a low noise call state (S2 → S5), and when PN is equal to or greater than PNth0 and less than PNth1, a noisy call state (S2 → S3 → S6), when the PN is equal to or greater than PNth1, it has a function of determining the call state under high noise (S2 → S3 → S4), and the determination result is the near end state 22. It is assumed that PNth0 and PNth1 can be set empirically depending on the usage environment.

  As shown in FIG. 3, the reception output limit estimation means 60 includes a reception output restriction reception input power calculation means 62, a low noise call state reception output upper limit level holding means 64, a noise call state reception output upper limit level holding means 65, a high A noisy call state reception output upper limit level holding unit 66, reception output upper limit level selection unit 67, and reception output limit loss amount update unit 69 are configured.

  The reception output limit reception input power calculation means 62 outputs the result of instantaneous value power calculation of the reception input signal 2 as reception output restriction reception input power 63. The reception output upper limit level selection unit 67 is held in the low noise call state reception output upper limit level holding unit 64 in the low noise call state with respect to the state of the near end state 22 output from the near end background noise estimation unit 21. The upper limit level held in the noisy call state reception output upper limit level holding means 65 in the noisy call state and the noisy call state reception output upper limit level holding means 66 in the noisy call state. The held upper limit levels are output as received output upper limit levels 68, respectively.

  The low noise call state reception output upper limit level holding means 64, the noise call state reception output upper limit level holding means 65, and the high noise call state reception output upper limit level holding means 66 so as to provide an appropriate speaker output for each state. In addition, it is possible to empirically set the reception output upper limit level according to the use environment.

  The reception output limit loss amount updating means 69 is the minimum because the result of multiplying the reception output limit reception amount 63 by the reception output limit loss amount 61 is less than the reception output upper limit level 68. A function of updating the received output limit loss amount 61 to be a large loss amount.

  As shown in FIG. 2, the voice switch unit 10 includes a received voice determination unit 30, a transmitted voice determination unit 40, a loss amount control unit 50, a second fixed loss amount holding unit 51, and a first fixed loss amount holding unit 52. Is done. As already described, the first fixed loss amount held in the first fixed loss amount holding means 52 is a fixed loss amount capable of suppressing a predetermined echo and howling. Further, the second fixed loss amount held in the second fixed loss amount holding means 51 is a loss amount smaller than the first fixed loss amount, and is a fixed loss amount capable of suppressing howling.

  The received voice determination unit 30 has a function of determining the presence or absence of a received voice by monitoring the received input signal (far-end voice) 2, and sets the received voice determination unit 30 as a received state 31.

  The transmission voice determination means 40 determines the presence or absence of transmission voice by monitoring the voice signal 15 after echo cancellation, and particularly indicates that the adaptive filter state 16 is in an unconverged state and the reception state 31 is reception voice. When it is judged, it is difficult to judge that the echo left behind by the echo canceller is transmitted speech (for example, if the near-end acoustic echo has a magnitude that cannot be ignored relative to the near-end talker speech, it is known in advance. The voice signal 15 after the echo cancellation is larger than the maximum near-end acoustic echo is determined to have a voice), and the judgment result of the transmission voice judgment unit 40 is a transmission state 41.

  In the background noise estimation means 21 described above, the long-term average power of the absolute value component of the audio signal 15 after echo cancellation is used as the estimated noise level PN. The long-time referred to here is the result of averaging the power. Non-speech indicates an appropriate near-end noise level, and even when there is speech, it is a time during which the influence on the near-end noise level is minimal, and the main speech level, near-end background noise, Adjust appropriately from the response speed that changes to each call state. Specifically, in this example, the adjustment was appropriately made in the range of 1 second to 5 seconds.

  For example, when comparing the long-term average power with only background noise lasting for a long time with the long-term average power with both background noise and voice, it is negligible for voice in a normal call There is a clear difference for a voice that is several tens of dB or more larger than the voice in a normal call.

  The loss amount control means 50 is a loss amount to be inserted by the reception loss insertion means 11 and the transmission loss insertion means 13 from the reception state 31, the transmission state 41, the estimated loss amount 23, the near end state 22, and the reception output limit loss amount 61. To control. The loss amount control performed by the loss amount control means 50 is shown in Table 1, and the control procedure of the loss amount control means 50 is shown in the flowchart of FIG. Note that the notation “x” in Table 1 indicates that the corresponding state does not exist.

  When the near-end state 22 is a low noise call state (S10 → S11), the reception state refers to the reception state 31, the transmission state refers to the transmission state 41, and in the reception voiced state and the transmission silent state, (S11 → S12), the received output limit loss amount is inserted into the received loss insertion means 11 and the estimated loss amount is inserted into the transmitted loss insertion means 13 as a loss. In the reception silence state and the transmission sound state, the reception sound state and the transmission sound state, the reception silence state and the transmission silence state (S11 → S13), the reception loss insertion means 11 receives the reception output limit loss amount or the estimated loss. The greater of the amount is inserted as a loss, and no loss is inserted into the transmission loss insertion means 13.

  When the near-end state 22 is a noisy call state (S10 → S14), the reception state refers to the reception state 31 and the transmission state refers to the transmission state 41. Since the noise level is high, the voice is always transmitted. Therefore, the possible states are either a received sound state and a transmitted sound state, or a received soundless state and a transmitted sound state. In the received sound state and the transmitted sound state (S14 → S15) ), The received output limit loss amount is inserted into the received loss insertion means 11, and the first fixed loss amount is inserted into the transmitted loss insertion means 13 as a loss. In the no-speech state and the transmission-sounding state (S14 → S17), the greater one of the reception output limit loss amount and the first fixed loss amount is inserted into the reception loss insertion unit 11 as a loss, and the transmission loss insertion unit 13 is inserted. Does not insert loss.

  When the near-end state 22 is a high-noise call state (S10 → S16), the reception state refers to the reception state 31 and the transmission state refers to the transmission state 41. Since the background noise level is high, there is always a voice transmission state. Therefore, a possible state is either a received sound state and a transmitted sound state, or a received silence state and a transmitted sound state. In any case, the received loss insertion means 11 receives a received output limit loss. The amount of loss is inserted, and the second fixed loss amount is inserted into the transmission loss insertion means 13 as loss.

  The embodiment of the present invention can be implemented as a program that, when installed in a general-purpose information processing device, causes the general-purpose information processing device to realize a function corresponding to the above-described voice communication device. The program is recorded on a recording medium and installed in the general-purpose information processing apparatus, or is installed in the general-purpose information processing apparatus via a communication line, whereby the above-described voice call is made to the general-purpose information processing apparatus. Functions corresponding to the device can be realized. The general-purpose information processing apparatus is, for example, a CPU (Central Processing Unit) such as a microcomputer chip mounted on a remote call device or the like.

  According to the present invention, it is possible to make a good call under various noise environments, and therefore, in an environment where background noise changes variously, for example, in a construction site or a bathroom or a building facing a busy road. A voice communication device suitable for use as a door phone or the like can be realized.

1 is an overall configuration diagram of a voice call device according to an embodiment of the present invention. The block block diagram of an audio | voice switch part. The block block diagram of a receiving output restriction | limiting estimation means. The flowchart which shows a state determination procedure. The flowchart which shows a loss insertion procedure.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Reception input terminal 2 Reception input signal 3 Transmission output terminal 4 Near-end signal 5 after transmission loss insertion Transmission input terminal 6 Reception output terminal 7 Transmission input signal 10 Voice switch part 11 Reception loss insertion means 12 After reception loss insertion Voice signal 13 Transmission loss insertion means 14 Echo canceller 15 Voice signal 16 after echo cancellation Adaptive filter state 20 Loss amount estimation means 21 Background noise estimation means 22 Near end state 23 Estimated loss amount 30 Received voice determination means 31 Reception state 40 Transmission voice determination means 41 Transmission state 50 Loss amount control means 51 Second fixed loss amount holding means 52 First fixed loss amount holding means 60 Received output limit estimation means 61 Received output limit loss amount 62 Received output limit received input power calculation Means 63 Received output restriction received input power 64 Low noise call state reception output upper limit level holding means 65 Noisy call state reception output upper limit level Holding means 66 High-noise conversation state reception output upper limit level holding means 67 Reception output upper limit level selection means 68 Reception output upper limit level 69 Reception output limit loss amount updating means

Claims (2)

One of two parties in different locations is a near-end speaker, the other is a far-end speaker, the end to which the voice signal of the near-end speaker is input is a transmission input end, and the far-end talk An end to which the voice signal of the far-end speaker is input, an end to which the voice signal of the far-end speaker is output is an output end, and an end to which the voice signal of the near-end speaker is output , A transmission signal path and a reception signal path between the transmission input terminal and the transmission output terminal and between the reception input terminal and the reception output terminal, respectively. Voice communication means to make a call between
This voice call means
Audio signal detection means for detecting the presence or absence of audio signals in the transmission signal path and the reception signal path, respectively;
Loss insertion means for adaptively setting a loss insertion location and a loss insertion amount of the transmission signal path or the reception signal path according to a detection result of the voice signal detection means;
An echo canceller provided between the reception output terminal and the transmission input terminal and canceling an echo generated by a voice signal of the far-end speaker;
In a voice communication device comprising: a loss amount estimating means for estimating a required loss insertion amount based on a voice signal after insertion loss reception and a voice signal after echo cancellation by the echo canceller;
An average signal strength at a predetermined time of a voice signal input from the transmission input terminal or a voice signal after echo cancellation by the echo canceller is set as an estimated noise level, and predetermined according to the strength of the estimated noise level. It includes a status classification means for classifying the quality of the call environment in a plurality of stages,
The state classification means is based on the intensity of the estimated noise level.
A low-noise call state in which the estimated value of the loss insertion amount by the loss amount estimation means is within the applicable range and the presence or absence of a voice signal by the voice signal detection means and echo cancellation by the echo canceller is possible,
A noisy call in which the estimated value of the loss insertion amount by the loss amount estimation means is out of the applicable range, the detection of the presence or absence of the voice signal by the voice signal detection means is uncertain, and echo cancellation by the echo canceller is possible Status,
Further, the call with high noise, the estimated value of the loss insertion amount by the loss amount estimation means is out of the applicable range, and the voice signal detection means cannot detect the presence or absence of the voice signal and the echo canceller cannot cancel the echo. Status
With the means to classify
The loss insertion means includes
It is classified into a call state under low noise by the state classification means,
When a voice signal is detected in the reception signal path and a voice signal is not detected in the transmission signal path, a predetermined reception output limit loss amount is inserted into the reception signal path as a loss, and the transmission signal path Insert the loss insertion amount estimated by the loss amount estimation means,
When a voice signal is detected in the transmission signal path and a voice signal is not detected in the reception signal path, or when a voice signal is detected in both the reception signal path and the transmission signal path, or the reception When a speech signal is not detected in both the signal path and the transmission signal path, the larger one of the received output limit loss amount or the loss insertion amount estimated by the loss amount estimation means is inserted into the received signal path as a loss insertion. And
Classified by the state classification means into a call state under noise,
When the voice signal is detected in the transmission signal path and the voice signal is not detected in the reception signal path, the reception output limit loss amount or a predetermined echo and howling can be suppressed in the reception signal path. Insert the larger of the fixed loss amount of
When a speech signal is detected in both the reception signal path and the transmission signal path, the reception output limit loss amount is inserted into the reception signal path as a loss, and the first fixed loss amount is added to the transmission signal path. Inserting loss,
It is classified into a call state under high noise by the state classification means,
When a voice signal is detected in the transmission signal path and a voice signal is not detected in the reception signal path, or when a voice signal is detected in both the reception signal path and the transmission signal path, the reception signal path A voice communication device characterized by inserting a loss of the received output limit loss amount into the transmission signal path and inserting a second fixed loss amount smaller than the first fixed loss amount into the transmission signal path . A program that, when installed in a general-purpose information processing apparatus, causes the general-purpose information processing apparatus to realize a function corresponding to the voice call device according to claim 1 .

Images