JP5812932B2 - Voice listening device, method and program thereof - Google Patents

Description

  The present invention relates to a voice listening device for acquiring a transmission signal and a reception signal from a telephone, a method thereof, and a program.

  2. Description of the Related Art Conventionally, there is a voice listening device that acquires a transmission signal and a reception signal of a telephone. As one of the voice listening devices, a voice recognition device is known that recognizes a transmission signal and a reception signal as voice and outputs listening data as a voice file for reproduction. In the speech recognition apparatus, when the speech signal and the reception signal are separately recognized, the crosstalk state in which the speaker on the transmission side and the speaker on the reception side speak simultaneously has been a problem. In the case of the crosstalk state, each utterance becomes a distorted signal unlike usual. The disturbed transmission signal and reception signal cause misrecognition of speech recognition.

  In order to prevent misrecognition of speech recognition in the crosstalk state, speech section detection is performed for each of the transmission signal and the reception signal, and only one of the speakers detects the speech state section and recognizes only that section. A speech recognition apparatus 110 is disclosed in Patent Document 1.

  The operation will be briefly described with reference to the functional configuration of the conventional speech recognition apparatus 110 shown in FIG. The voice recognition device 110 includes a side tone suppression processing unit 21, a transmission voice segment detection unit 22, a reception voice segment detection unit 23, a voice segment information management unit 24, a transmission signal extraction unit 25, a reception signal extraction unit 26, and a transmission. A signal recording unit 16, an incoming signal recording unit 17, and a speech recognition processing unit 111 are provided.

  The side tone suppression processing unit 21 removes the side tone signal that wraps around the received signal. A side sound is a transmission signal on the terminal side that is output in a small amount to a speaker on the terminal side for the purpose of facilitating conversation. The side sound signal is added to the received signal in the side sound circuit 15 in the telephone 19.

  The transmission voice section detection unit 22 detects a voice section and a non-voice section of the transmission signal. The received voice section detector 23 detects a voice section and a non-voice section of the received signal. The voice section information management unit 24 inputs the voice section and the non-voice section of the transmission signal and the reception signal, and specifies the transmission voice extraction section and the reception voice section that are not in the crosstalk state.

  The voice recognition processing unit 111 performs voice recognition processing on the transmission signal in the transmission voice extraction section and the reception signal in the reception voice extraction section to generate listening data and to output and reproduce the voice file. .

JP 2006-343642 A

  In the conventional speech recognition apparatus 110 that recognizes a transmission / reception signal for the purpose of listening to a voice, there is a problem that a reception signal in a crosstalk state cannot be extracted.

  The present invention has been made in view of such a problem, and an object of the present invention is to provide an audio listening device, a method thereof, and a program capable of extracting a reception signal in a crosstalk state.

  The voice listening device of the present invention includes a side-tone suppression processing unit, a received voice segment detecting unit, a received signal extracting unit, and a transmitted / received signal recording unit. The side tone suppression processing unit outputs a reception signal after side tone suppression in which the side tone signal is suppressed by receiving the reception signal and the transmission signal from the telephone. The reception voice section detection unit receives the side-tone-suppressed reception signal, detects a voice section of the side-tone-suppressed reception signal, and outputs reception voice section information. The received signal extraction unit receives the received voice segment information and the received signal, and outputs a received signal corresponding to the received voice segment information as a side tone removed received signal. The transmission / reception signal recording unit records the side-tone-removed reception signal and the transmission signal.

  According to the voice listening device of the present invention, the received voice section is detected from the side-tone-suppressed received signal in which the side-tone signal is suppressed, and the received signal in the received voice section is recorded as the received signal. Even in this case, it is possible to listen to all received signals without omission.

The figure which shows the function structural example of the audio | voice listening apparatus 100 of this invention. The figure which shows the operation | movement flow of the audio | voice listening apparatus 100. The figure which shows the function structural example of the audio | voice listening apparatus 200 of this invention. The figure which shows the function structural example of the audio | voice listening apparatus 300 of this invention. The figure which shows the function structural example of the received signal extraction part 301. FIG. The figure which shows the example of the side tone removal received signal which the received signal extraction part 301 outputs. The figure which shows the function structural example of the audio | voice listening apparatus 400 of this invention. The figure which shows the function structural example of the audio | voice listening apparatus 500 of this invention. The figure which shows the function structural example of the receiving area extraction part 503. FIG. The figure which shows the operation | movement flow of the received signal extraction part 503. The figure which shows the example of the side tone removal received signal which the received signal extraction part 503 outputs. The figure which shows the function structural example of the conventional speech recognition apparatus 110. FIG.

  Embodiments of the present invention will be described below with reference to the drawings. The same reference numerals are given to the same components in a plurality of drawings, and the description will not be repeated.

  FIG. 1 shows an example of the functional configuration of a voice listening device 100 according to the present invention. The operation flow is shown in FIG. The voice listening device 100 includes a side-tone suppression processing unit 10, a received voice section detecting unit 20, a received signal extracting unit 30, and a transmitted / received signal recording unit 40. The audio listening device 100 is realized by a predetermined program being read into a computer composed of, for example, a ROM, a RAM, and a CPU, and the CPU executing the program. The same applies to the audio listening devices shown in other embodiments described below.

  The side sound suppression processing unit 10 receives the reception signal and the transmission signal from the telephone 19 and outputs a side sound suppressed reception signal with the side sound signal suppressed (step S10). A telephone set 19 including a microphone 11, a speaker 12, a transmission unit 13, a reception unit 14, and a side sound circuit 15 is the same as that shown in the related art. Note that reference numerals of the telephone 19 are omitted for convenience of drawing. The telephone 19 is a general one, and the reception signal is the other party's speaker voice, and the transmission signal is the other party's voice.

  Side sound suppression processing unit 10 receives the received signal and the transmitted signal, suppresses the transmitted signal superimposed on the received signal, and outputs the received signal with the side sound suppressed. The side tone suppression processing unit 10 can be configured by a general echo canceller. The side-tone-suppressed reception signal in this example is a signal in which the utterance voice of the sender superimposed on the utterance voice of the other party is suppressed.

  The received voice section detection unit 20 receives the side-tone-suppressed received signal, detects the voice section of the side-tone-suppressed received signal, and outputs received voice section information (step S20). The received voice section information is section information representing the received voice section. As a method for detecting a voice section, a method based on a general sound volume (power) may be used. Alternatively, speech segment detection using a speech model and a non-speech model based on a mixed normal distribution (GMM: Gaussian Mixture Model) may be performed. In the case of voice segment detection using a mixed normal distribution, the voice / non-voice discrimination performance can be improved by using the above-described side-tone-suppressed reception signal for GMM learning.

  The reception signal extraction unit 30 receives the reception voice section information and the reception signal output from the reception voice section detection unit 20, and outputs a reception signal corresponding to the reception voice section information as a side tone removal reception signal (step S30). ). In addition, the reception signal extraction unit 30 sets the amplitude of the side-tone-removed reception signal other than the reception voice section information to 0 (silence) and outputs it. The side-tone-removed reception signal is a signal only for the reception signal in the section corresponding to the reception section detected as the reception voice section by the reception voice section detection unit 20.

  Since the side-tone-removed reception signal is a reception signal corresponding to the reception voice section information detected from the voice signal of the listener who suppresses the voice of the sender, the voice signal (reception signal) of the speaker in the crosstalk state is also used. Extract without leakage. That is, it is possible to listen to the received signal even in a crosstalk state where the received signal and the transmitted signal overlap.

  The transmission / reception signal recording unit 40 records the side-tone-removed reception signal and the transmission signal output from the reception signal extraction unit 30 (step S40). The transmission / reception signal recording unit 40 generates a transmission signal file and an audio file of a side-tone-removed reception signal synchronized in time with the transmission signal. The audio files may be two separate files or one stereo file.

  As described above, according to the voice listening device 100, the received voice section information is detected from the side-tone-suppressed received signal in which the side-tone signal is suppressed, and the received signal corresponding to the received voice-section information is detected as the side-tone-removed received signal. Therefore, it is possible to listen to all received signals without omission even in a crosstalk state. The side-tone suppression processing unit 10 suppresses the side-tone only for the purpose of detecting the reception voice section, so that the amount of suppression can be increased. In other words, since the waveform may be distorted, it is possible to detect an accurate received voice interval. The received signal extracted corresponding to the accurate received voice section information can be a signal with little distortion, and thus is a signal that is easy to listen to and easy to use for voice recognition and the like.

  As shown in FIG. 1, the voice recognition processing unit 50 is indicated by a broken line, and the side recognition removed reception signal and the transmission signal recorded in the transmission / reception signal recording unit 40 are subjected to voice recognition processing by the voice recognition processing unit 50. You may convert into text data. In addition, the audio viewing unit 60 may be configured to associate the text data of the speech recognition result, the side-tone-removed reception signal, and the audio file of the transmission signal with each other so that the audio file can be efficiently viewed. The audio viewing unit 60 is, for example, a functional unit that includes application software that displays text data on a display device and that can reproduce an audio file.

  Next, a voice listening device 200 that extracts a reception signal from a side tone suppression reception signal will be described.

  FIG. 3 shows a functional configuration example of the voice listening device 200 of the present invention. The audio listening device 200 is different from the audio listening device 100 in that it further includes a side tone low suppression processing unit 201. The side tone low suppression processing unit 201 receives the reception signal and the transmission signal as inputs, and outputs a low suppression amount side reception suppressed signal having a large original sound addition rate to the reception signal extraction unit 30.

  The reception signal extraction unit 30 receives the reception voice section information output from the reception voice section detection unit 20 and the reception signal with the side-tone low-suppressed input, and receives the side-tone-suppressed reception signal corresponding to the reception voice section information. Output as a sidetone-removed reception signal.

  The echo suppression amount of the side tone suppression processing unit 201 is set lower than that of the side tone suppression processing unit 10. The side-tone suppression processing unit 10 sets the echo suppression amount with a high original sound addition rate A (for example, A = 0.1) with respect to the echo suppression amount. If the echo suppression amount is set high, the reception signal is distorted, but the reception voice section information can be accurately detected.

  For example, the distortion of the side-tone-suppressed received signal output from the side-tone low suppression processing unit 201 that is set to a low echo suppression amount of the original sound addition rate A = 0.6 is less distortion that the transmission signal is suppressed. This is a received signal. Since the reception signal extraction unit 30 extracts the side tone removal reception signal from the reception voice section information and the reception signal with suppressed sidetone low suppression with less distortion, even if the side sound removal reception signal is subjected to voice recognition processing, for example. An audio signal with few misrecognitions can be obtained.

  As described above, according to the voice listening device 200, the received voice section detecting unit 20 can increase the echo suppression amount of the voice signal for detecting the received voice section information so that accurate received voice section information can be detected. A side-tone-removed received signal with less distortion can be extracted from a received signal with reduced side-tone low suppression corresponding to accurate received voice section information.

  Next, a description will be given of the voice listening apparatus 300 in which noise is superimposed on the side sound removal received signal other than the received voice section where the amplitude of the received signal is zero.

  FIG. 4 shows a functional configuration example of the voice listening device 300 of the present invention. The voice listening device 300 differs from the voice listening device 100 in the operation of the received signal extraction unit 301. The received signal extraction unit 301 operates so as to superimpose noise with a small amplitude (for example, white noise) on the place where the amplitude of the above-described side-tone-removed received signal is 0 (silence).

  The reception signal extraction unit 301 outputs the reception signal corresponding to the reception voice section information or the reception signal after the side-tone low suppression as the side sound removal reception signal, and the side-tone removal reception signal of the section not corresponding to the reception voice section information. , Output as low noise level.

  Since the amplitude of the side-tone-removed reception signal does not change abruptly in the audio listening device 300 compared to the audio listening devices 100 and 200 described above, the auditory continuous listening effect (for example, http://www.brl.ntt. co.jp/IllusionForum/a/continuityIllusion/en/index.html), which is easier to work with, and can prevent the listener from making mistakes. The voice listening device 300 not only prevents the listener from hearing errors but also improves the stability of voice recognition accuracy.

  FIG. 5 shows a functional configuration example of the received signal extraction unit 301. The received signal extraction unit 301 includes noise generation means 3010 and noise superimposition means 3011. The noise generation unit 3010 generates white noise having a very small volume level of, for example, about −80 dB. The white noise can be easily generated by a conventional method using normal random numbers, for example.

  The noise superimposing means 3011 receives the reception voice section information and the reception signal or the reception signal with the side-tone low suppressed, and sets the section corresponding to the reception voice section information as the reception signal or the side-tone low-suppression reception signal. A sidetone-removed reception signal with white noise in the section is output.

  FIG. 6 shows an example of the sidetone-removed reception signal output by the reception signal extraction unit 301. The first line is a reception signal, and the second line is reception voice section information output from the reception voice section detection unit 20 with the reception signal as an input. The third line is white noise output from the noise generating unit 3010. The fourth line shows a side-tone-removed reception signal output by the noise superimposing means 3011 when the above-described reception voice section information, reception signal, and white noise are input. It can be seen that the reception signal is output in the section corresponding to the reception voice section information, and white noise is output in the other sections.

  Note that when the volume level of the received signal is low, there is a problem that the speech recognition rate is reduced due to noise superimposition. Therefore, the volume level of the received signal or the level of noise to be superimposed may be changed according to the S / N ratio of the received signal. In that case, S / N ratio detection means 3012 for detecting the S / N ratio of the reception signal or the reception signal with suppressed side tone low is provided in the reception signal extraction unit 301, and the S detected by the S / N ratio detection means 3012. The amplitude of the white noise generated by the white noise generating unit 3010 according to the / N ratio may be automatically controlled so that the S / N ratio becomes, for example, 30 dB or more. The S / N ratio detection means 3012 may be replaced with a power detection means for detecting the power of the reception signal or the reception signal with the side-tone low suppressed. In addition, although Example 3 demonstrated the example which superimposes white noise on a side-tone removal receiving signal, noise does not need to be limited to white noise. For example, pink noise whose power spectral density is inversely proportional to the frequency may be used. Even noise other than white noise and pink noise may be used that exhibits the same effect.

  Further, in the above description, an example in which a side-tone-removed reception signal in a section that does not correspond to the reception voice section information is output as noise with a low sound volume level has been described. Therefore, noise with a low volume level may be superimposed on the entire received voice including the section corresponding to the received voice section information. Similarly, an auditory continuous listening effect can be obtained.

  Next, the voice listening device 400 in which the received signal extraction unit operates so as to reduce the amplitude of the side-tone-removed received signal in a section where the volume level on the transmitting side is excessively high will be described.

  FIG. 7 shows a functional configuration example of the audio listening device 400. The voice listening device 400 differs from the voice listening device 100 in that it includes a transmitted voice section detecting unit 401 and an operation of the received signal extracting unit 402.

  The transmission voice section detection unit 401 receives a transmission signal, detects a voice section of the transmission signal, detects a transmission voice section, detects a volume level in the transmission voice section, and detects the volume level. And the transmission voice section are output. The method for detecting the voice segment is the same as that of the received voice segment detector 20 described above.

  The reception signal extraction unit 402 refers to the volume level output from the transmission voice section detection unit, and controls the amplitude of the reception signal corresponding to the section when the volume level is a predetermined value or more. For example, when the volume level of the transmission signal is −20 dB or more, the predetermined value is when the ratio of the volume level of the transmission signal / the maximum volume level (maximum value of the volume level that can be handled by the audio listening device 400) is α, For example, the amplitude of the side-tone-removed reception signal including the transmission voice section is reduced by (1−α) times.

  By doing in this way, the volume level of the side-tone-removed reception signal in the section where the volume level on the transmission side is excessively high can be lowered, and the sneak sound of the transmission signal can be suppressed from being heard at a high volume. I can do it. The idea of the audio listening device 400 may be combined with the audio listening devices 200 and 300, and even in that case, the same effect can be obtained.

  Next, a description will be given of a voice listening device 500 that outputs a received signal corresponding to a received voice section with a margin time before and after the received voice section information as a side-tone-removed received signal.

  FIG. 8 shows a functional configuration example of the audio listening device 500. The voice listening device 500 includes a side-tone suppression processing unit 10, a received transmission signal recording unit 501, a voice segment detection unit 502, a received signal extraction unit 503, and a side-tone-removed reception signal recording unit 504. . The side-tone suppression processing unit 10 is the same as the audio listening device 100 as is clear from the reference numerals.

  Sidetone suppression processing unit 10 receives a reception signal and a transmission signal from a telephone (not shown), and outputs a sidetone-suppressed reception signal in which the sidetone signal is suppressed.

  The received transmission signal recording unit 501 records the side-tone-suppressed reception signal output from the side-tone suppression processing unit 10 and the transmission signal. The audio listening device 500 does not perform real-time processing. For example, the audio listening device 500 reproduces the audio of the audio file in which the conversation between the receiver and the transmitter is recorded, and uses the side-signal-suppressed reception signal and the transmission signal that have been recorded once. Process.

The speech section detection unit 502 reads the side-tone-suppressed reception signal and the transmission signal from the reception-speech signal recording unit 501, and the start times Ru _{1 to} Ru _n of the plurality of speech sections of the side-tone-suppressed reception signal end time _Rd 1 ~ Rd _n, and output as speech segment information by detecting the start time _Su 1 to SU _n and the end time _Sd 1 to SD _n of a plurality of speech segment of said transmission signal.

The received signal extraction unit 503 receives the speech section information Ru _{1 to} Ru _n , Rd _{1 to} Rd _n , Su _{1 to} Su _n , Sd _{1 to} Sd _n and the received signal, and receives the speech section of the side-tone-suppressed received signal. A reception signal corresponding to a reception voice section to which a margin time is given before and after the information is output as a side-tone-removed reception signal.

  The side tone removed received signal recording unit 504 records the side tone removed received signal output by the received signal extracting unit 503.

  According to the voice listening device 500, margin time can be given before and after the sidetone removal received signal, so that, for example, a voice section is lost when the sidetone removal received signal starts with a consonant. It is possible to prevent erroneous detection.

FIG. 9 shows a more specific functional configuration example of the received signal extraction unit 503. The received signal extraction unit 503 includes a margin time giving unit 512 and a side tone removal received signal extracting unit 522. Margin time imparting means 512, the speech section information _{Ru 1 ~Ru n, Rd 1 ~Rd} n, Su 1 ~Su n, as input Sd 1 to SD _n, the start time of the speech section of the sidetone suppression already received signal Ru within ₁ ～Ru predetermined time than _n Rms of past time, it determines whether the end time _Sd 1 to SD _n speech period of the transmission signal is present. Then, in the end time _Rd 1 ~ Rd _n predetermined time than Rme future time of the audio section of the sidetone suppression already received signal, whether the start time _Su 1 to SU _n in the speech section of the transmission signal is present Determine. When the disclosure time and the end time do not exist within the predetermined time, a margin time of the predetermined time is given before and after the reception voice section information.

Further description will be given with reference to the operation flow of the voice section detection unit 502 and the reception signal extraction unit 503 shown in FIG. The voice section detection unit 502 reads the side-tone-suppressed reception signal and the transmission signal from the reception-speech signal recording unit 501, and starts the side-tone-suppressed reception signal Ru _{1 to} Ru _n and the end time Rd ₁ to detecting the Rd _n (step S5020). The start time is the rise time of the received voice section, and the end time is the fall time of the received voice section. Similarly, to detect an end time _Sd 1 to SD _n and start time _Su 1 to SU _n of the transmission voice section (Step S5021). FIG. 11 illustrates two pieces of received voice section information Ru ₁ and Ru ₂ , Rd ₁ and Rd ₂ , and two pieces of transmitted voice section information Su ₁ and Su ₂ , Sd ₁ and Sd ₂ . Numbers are assigned to the start time and end time in order of elapsed time.

The received signal extraction unit 503 initializes (i = 1) a variable i representing the number of the voice section (step S5121). Then, it is determined whether or not the end time Sd ₁ of the transmission signal exists within the time Rms (for example, 0.5 seconds) past the start time Ru ₁ of the first voice section of the reception signal with sidetone suppression. (Step S5122). The Rms seconds in past time, when the end time Sd ₁ of the transmission signal does not exist (No in step S5122), the start time Ru ₁ of the first speech section of the sidetone suppression already received signal _{Ru 1} = _Ru 1 - Rms seconds are determined (step S5123). The Rms seconds within a time in the past, (Yes in Step S5122) If the end time Sd ₁ of the transmission signal is present, the start time Ru ₁ of the first speech section of the sidetone suppression already received signal, the first transmission signal End time Sd ₁ = Sd ₁ second is determined (step S5124). Here, for example, Rms = 0.5 seconds and Rme = 0.1 seconds, but other predetermined times may be used. The reason why the time widths of Rms and Rme are changed is that it is advantageous to secure a long margin time for the purpose of dealing with a case where the sound volume starts small and is difficult to detect as a voice section at the start of utterance. It depends. On the other hand, in the case of Japanese ending with vowels, there is little inconvenience even if the margin time of the end time of the received signal is short.

Next, the sidetone suppression already received signals of the first end time Rd ₁ in more Rme of seconds in the future time of the speech section, determines whether the start time Su ₂ of the transmission signal is present (step S5125). In RME of seconds in the future time, when the start time Su ₂ of the transmission signal does not exist (No in step S5125), the sidetone suppression already received signals of the first end time Rd ₁ speech period _Rd 1 _{= Rd} 1 + Rme Second is determined (step S5126). In Rme seconds in the future time, (Yes in step S5125) When the start time Su ₂ of the transmission signal is present, the end time Rd ₁ of the first speech section of the sidetone suppression already received signal, the start of the transmission signal Time Su ₂ (Rd ₁ = Su ₂ ) seconds is determined (step S5127).

  The above processing is continued while incrementing the variable i representing the number of the voice section until the voice section is completed (No loop in step S5129). That is, a margin time of a predetermined time width is given before and after the voice section of the reception signal with side-tone suppressed.

  FIG. 11 shows an example of a side-tone-removed reception signal that has been listened to by the voice listening apparatus 500. The first line is the received voice section information of the side-tone-suppressed received signal, the second line is the transmitted signal, the third line is the transmitted voice section information of the transmitted signal, and the fourth line is output from the margin time giving means 512. Received speech section information with margin time added, and the fifth line is a sidetone-removed received signal output from the received signal extraction unit 503. A margin time of a predetermined time is added to the output signal of the margin time giving means 512 before and after the reception voice section information as indicated by a broken line. The speech waveform of the side-tone-removed reception signal corresponding to the reception speech section information to which the margin time has been added is expanded in the front-rear direction of time as compared with that shown in FIG.

  As described above, the voice listening device 500 can extract and record a received signal corresponding to the received voice section information to which a predetermined margin time is given as a side-tone-removed received signal. The functional configuration of the audio listening device 500 is not limited to the example shown in FIG. For example, although the description has been made with the example in which the voice section of the transmission signal and the reception signal is detected by the voice section detection unit 502, the voice section of the reception signal is performed by the reception voice section detection unit 20, and the voice section of the transmission signal is performed. As described in the fourth example, the transmission voice section detection unit 401 may perform the voice section information from each. In addition, although an example in which the received transmission signal recording unit 501 is provided has been described, the voice data before the voice acquisition process is recorded in the transmission / reception signal recording unit 40 described in the first to fourth embodiments. May be. Thus, the functional configuration of the fifth embodiment is not limited to the functional configuration illustrated in FIG.

  Note that the idea of giving a margin time having a predetermined time width before and after the speech section of the reception signal after side-tone suppression can be combined with the second to fourth embodiments.

  According to the voice listening device described in the above embodiment, the received voice section information is detected from the side-tone-suppressed received signal in which the side-tone signal is suppressed, and the received signal corresponding to the received voice-section information is detected as the side-tone-removed received signal. Therefore, it is possible to listen to all received signals without omission even in a crosstalk state.

  Note that the processes described in the above method and apparatus are not only executed in time series according to the order of description, but may also be executed in parallel or individually as required by the processing capability of the apparatus that executes the processes. Good.

  When the processing means in the above apparatus is realized by a computer, the processing contents of the functions that each apparatus should have are described by a program. Then, by executing this program on the computer, the processing means in each apparatus is realized on the computer.

  The program describing the processing contents can be recorded on a computer-readable recording medium. As the computer-readable recording medium, for example, any recording medium such as a magnetic recording device, an optical disk, a magneto-optical recording medium, and a semiconductor memory may be used. Specifically, for example, as a magnetic recording device, a hard disk device, a flexible disk, a magnetic tape or the like, and as an optical disk, a DVD (Digital Versatile Disc), a DVD-RAM (Random Access Memory), a CD-ROM (Compact Disc Read Only) Memory), CD-R (Recordable) / RW (ReWritable), etc., magneto-optical recording media, MO (Magneto Optical disc), etc., semiconductor memory, EEP-ROM (Electronically Erasable and Programmable-Read Only Memory), etc. Can be used.

  This program is distributed by selling, transferring, or lending a portable recording medium such as a DVD or CD-ROM in which the program is recorded. Further, the program may be distributed by storing the program in a recording device of a server computer and transferring the program from the server computer to another computer via a network.

  Each means may be configured by executing a predetermined program on a computer, or at least a part of these processing contents may be realized by hardware.

Claims (9)

A side-tone suppression processing unit that receives a reception signal x ₁ and a transmission signal from a telephone and outputs a side-tone-suppressed reception signal x ₂ that suppresses the side-tone signal;
As input said side sound suppression already received signal x _2, the received voice section detection unit for outputting a received voice segment information by detecting a voice section of the side sound suppression already received signal x _2,
With the received signal x ₁ and the transmitted signal as inputs, a side tone low suppression processing unit that outputs a side suppression low-suppressed received signal x ₄ having a low suppression amount with a large original sound addition rate ;
As inputs the received voice segment information and the side tone low suppression already received signal x _4, regardless of whether the cross-talk state, the sidetone low suppression already received signal x ₄ corresponding to the received voice segment information, side A received signal extraction unit that outputs as a sound removal received signal x ₃ ;
The side-tone-removed reception signal x ₃ and the transmission / reception signal recording unit for recording the transmission signal,
A voice listening device comprising: A side-tone suppression processing unit that receives a reception signal x ₁ and a transmission signal from a telephone and outputs a side-tone-suppressed reception signal x ₂ that suppresses the side-tone signal;
As input said side sound suppression already received signal x _2, the received voice section detection unit for outputting a received voice segment information by detecting a voice section of the side sound suppression already received signal x _2,
As inputs the received voice segment information and the received signals x _1, regardless of whether the cross-talk state, the reception signal x ₁ corresponding to the received voice segment information, and outputs as a sidetone removed received signal x ₃ , and the reception signal extraction section for outputting a side tone removal received signal x ₃ sections that do not correspond to the received voice segment information as a small noise-volume level,
The side-tone-removed reception signal x ₃ and the transmission / reception signal recording unit for recording the transmission signal,
A voice listening device comprising: A side-tone suppression processing unit that receives a reception signal x ₁ and a transmission signal from a telephone and outputs a side-tone-suppressed reception signal x ₂ that suppresses the side-tone signal;
As input said side sound suppression already received signal x _2, the received voice section detection unit for outputting a received voice segment information by detecting a voice section of the side sound suppression already received signal x _2,
Detecting the voice section of the transmission signal to detect the transmission voice section, detecting the volume level in the transmission voice section, and outputting the transmission voice section and the volume level of the transmission voice section A transmission voice section detection unit;
With the received voice section information and the received signal x ₁ as inputs , regardless of whether or not in a crosstalk state , the received signal x ₁ corresponding to the received voice section information is output as a side-tone-removed received signal x ₃ , When the volume level is higher than a predetermined value with reference to the volume level output by the transmission voice section detection unit , a received signal extraction unit that controls to reduce the amplitude of the received signal x ₁ corresponding to the section ;
The side-tone-removed reception signal x ₃ and the transmission / reception signal recording unit for recording the transmission signal,
A voice listening device comprising: A side-tone suppression processing unit that receives a reception signal x ₁ and a transmission signal from a telephone and outputs a side-tone-suppressed reception signal x ₂ that suppresses the side-tone signal;
The side-tone-suppressed reception signal x ₂ and the reception transmission signal recording unit for recording the transmission signal,
It reads and the side tone suppression already received signal x ₂ and the transmission signal from the received transmission signal recording unit, and start time Ru ₁ ~Ru _n of a plurality of speech segment of said side sound suppression already received signal x ₂ a voice section detection unit for outputting a speech segment information end time _Rd 1 ~ Rd _n, and detects the start time _Su 1 to SU _n and the end time _Sd 1 to SD _n of a plurality of speech segment of said transmission signal,
With the voice segment information and the received signal x ₁ as inputs , regardless of whether or not a crosstalk state is present, the received voice segment with margin time added before and after the voice segment information of the side signal suppressed received signal x ₂ A received signal extraction unit that outputs a corresponding received signal x ₁ as a side-tone-removed received signal x ₃ ;
A side-tone-removed reception signal recording unit for recording the side-tone-removed reception signal x ₃ ;
A voice listening device comprising: Side-tone suppression processing process for receiving side-tone-suppressed reception signal x ₂ with side-tone signal being suppressed, with reception signal x ₁ and transmission signal from the phone as inputs,
With the side-tone-suppressed reception signal x ₂ as an input, a reception voice-section detection process of detecting a voice section of the side-tone-suppressed reception signal x ₂ and outputting reception voice section information;
With the received signal x ₁ and the transmitted signal as inputs, a side-tone low suppression process for outputting a side-tone low-suppressed received signal x ₄ with a low suppression amount with a large original sound addition rate ,
As inputs the received voice segment information and the side tone low suppression already received signal x _4, regardless of whether the cross-talk state, the sidetone low suppression already received signal x ₄ corresponding to the received voice segment information, side The process of extracting the received signal to be output as the sound-removed received signal x ₃ ;
The side-tone-removed reception signal x ₃ and the transmission / reception signal recording process for recording the transmission signal,
A voice listening method comprising: Side-tone suppression processing process for receiving side-tone-suppressed reception signal x ₂ with side-tone signal being suppressed, with reception signal x ₁ and transmission signal from the phone as inputs,
With the side-tone-suppressed reception signal x ₂ as an input, a reception voice-section detection process of detecting a voice section of the side-tone-suppressed reception signal x ₂ and outputting reception voice section information;
As inputs the received voice segment information and the received signals x _1, regardless of whether the cross-talk state, the reception signal x ₁ corresponding to the received voice segment information, and outputs as a sidetone removed received signal x ₃ , A reception signal extraction process for outputting the side-tone-removed reception signal x ₃ of the section not corresponding to the reception voice section information as noise having a low volume level ;
The side-tone-removed reception signal x ₃ and the transmission / reception signal recording process for recording the transmission signal,
A voice listening method comprising: Side-tone suppression processing process for receiving side-tone-suppressed reception signal x ₂ with side-tone signal being suppressed, with reception signal x ₁ and transmission signal from the phone as inputs,
With the side-tone-suppressed reception signal x ₂ as an input, a reception voice-section detection process of detecting a voice section of the side-tone-suppressed reception signal x ₂ and outputting reception voice section information;
Detecting the voice section of the transmission signal to detect the transmission voice section, detecting the volume level in the transmission voice section, and outputting the transmission voice section and the volume level of the transmission voice section The process of detecting the transmitted voice interval;
With the received voice section information and the received signal x ₁ as inputs , regardless of whether or not in a crosstalk state , the received signal x ₁ corresponding to the received voice section information is output as a side-tone-removed received signal x ₃ , When the volume level is greater than or equal to a predetermined value with reference to the volume level output by the transmission voice section detection process , a received signal extraction process for controlling to reduce the amplitude of the received signal x ₁ corresponding to the section ;
The side-tone-removed reception signal x ₃ and the transmission / reception signal recording process for recording the transmission signal,
A voice listening method comprising:   Answer signal from phone x ₁ Side-signal suppressed received signal x with side-tone signal suppressed ₂ Side-tone suppression process that outputs
  Received signal x with side-tone suppressed ₂ And the received transmission signal recording process for recording the transmission signal, and
  From the received speech transmission signal recording process, the side signal suppressed reception signal x ₂ And the above transmission signal, and the side-tone-suppressed reception signal x ₂ Start times Ru of multiple voice segments ₁ ~ Ru _n And end time Rd ₁ ~ Rd _n , And start times Su of a plurality of voice sections of the transmission signal ₁ ~ Su _n And end time Sd ₁ ~ Sd _n Detecting a voice segment and outputting it as voice segment information;
  Voice section information and received signal x ₁ , And the side-tone-suppressed received signal x regardless of whether or not it is in the crosstalk state. ₂ Received signal x corresponding to the received voice section with margin time before and after the voice section information of ₁ , Sidetone elimination received signal x _Three Receiving signal extraction process to be output as
  Above sidetone elimination received signal x _Three Side-tone-removed received signal recording process,
  A voice listening method comprising: A program for causing a computer to function as the audio listening device according to any one of claims 1 to 4 .

Images