JP2007221574A - Voice processing apparatus, voice processing method, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To produce voice data for well recording only a voice of a speaker when a plurality of microphones are disposed. <P>SOLUTION: An apparatus includes a first memory 21 for storing original voice data collected by at least more than two microphones after adding an identifier, an identifier extracting unit 42c for extracting the identifier added to the original voice data in which a level of the original voice data stored in the first memory 21 exceeds a predetermined threshold, and a voice data addition control unit 42d for adding the original voice data corresponding to what is except for the extracted identifier from among the original voice data read from the first memory 21 after attenuating. The voice data performed the addition process are to be produced. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a voice processing apparatus, a voice processing method, and a program that are suitable for application when, for example, a voice of a speaker is recorded in a conference.

  Conventionally, in a conference or the like, various techniques have been provided for recording a plurality of utterances uttered by spatially dispersed speakers using a plurality of microphones (also referred to as microphones in the following description). For example, there has been a technique of selecting (switching) and recording only a microphone signal placed near a speaker by the operation of the speaker or an operator. There has also been a technique for recording by automatically detecting the amplitude of the audio signal and selecting the speaker. Furthermore, there has been a technique for recording by adding (mixing) audio signals collected by all microphones. In recent years, the required storage capacity can be ensured even for long-time recording due to the increase in capacity and price of storage media such as hard disk drives and flash memories.

Patent Document 1 describes a data transmission system that collects a speaker's voice using a plurality of microphones.
Japanese Patent Laying-Open No. 2005-117134 (FIG. 14)

  By the way, the conventional technique of selecting and recording a speaker's microphone requires the on / off switching of the microphone selection operation. Therefore, there is a problem that the operation itself is troublesome and recording cannot be performed if the operation is wrong. For example, if the selection operation is forgotten to be switched on, voice cannot be recorded. Also, if you forget to switch off, unnecessary recording will continue.

  Further, in the technique of automatically recording by detecting the amplitude of the audio signal, the microphone is turned on after the audio signal becomes large to some extent, so that the speech start voice is cut off and recorded. Also, the microphone may be turned off when the voice becomes low just before the end of the speech, and the ending of the recorded speech is suddenly cut off. Or even if it is unnecessary noise, if it exceeds a certain threshold, the microphone is turned on and recorded. If it is easy to turn on the microphone so as not to cut off the head, the microphone is turned on even with noise. On the other hand, if it is set not to be turned on by noise, even if speech is started, it is regarded as noise and the microphone is not turned on, so the speech start voice is cut off. As described above, there is a possibility that the sound cannot be reliably recorded by the automatic recording technique.

  In addition, in the technique of recording by adding the audio signals collected by all microphones, unnecessary background noise collected by non-speaker microphones is added, resulting in a very high S / N (Signal to Noise). It gets worse. For example, when voice is recorded using 20 microphones, noise is added for 19 microphones to the voice of one speaker, and it cannot be said that the quality of the recorded voice is good.

  As described above, conventionally, there has been no technique for recording a good recording state from audio signals collected by a large number of microphones. In order to collect sound from a large number of microphones, a mixer is necessary, and thus an apparatus for efficiently creating a sound minutes has not yet been put into practical use.

  The present invention has been made in view of such circumstances, and an object of the present invention is to satisfactorily record the voice of a speaker when collecting sound with a large number of microphones.

  The present invention assigns an identifier to original voice data collected by at least two or more microphones and stores the identifier, and the identifier assigned to the original voice data in which the level of the stored original voice data exceeds a preset threshold value Are extracted and the original audio data corresponding to the extracted identifier other than the extracted identifier is attenuated and added.

  By doing in this way, it became possible to record the sound which extracted the speaker and attenuated the sound other than a speaker.

  According to the present invention, it is possible to extract a speaker and create a voice minutes in which a voice other than the speaker is attenuated. Therefore, the voice of the speaker is emphasized and unnecessary ambient noise is reduced. There is.

  Hereinafter, an embodiment of the present invention will be described with reference to the accompanying drawings. In the present embodiment, an example will be described in which the present invention is applied to a voice minutes creation apparatus that can create a voice minutes by collecting voices of speakers with a plurality of microphones. The audio minutes creation device used in this example identifies the speaker's microphone from the sound collected by the microphone, and attenuates the sound collected by the other microphones to emphasize and record the speaker's voice. Can create audio minutes.

  First, an example of the external configuration of the audio minutes creating apparatus of this example will be described with reference to FIG. FIG. 1 is a diagram showing a connection configuration example of a voice minutes creation system in which the voice minutes creation apparatus of this example is connected to various devices. The audio minutes creation apparatus 1 that creates the audio minutes of the present example collects the voice of the speaker from 26 microphones M1 to M26 as the original audio signal. Microphones M <b> 1 to M <b> 26 are provided with microphone numbers as identifiers for identification, and are installed on the conference table 5. The analog audio signals collected by the microphones M1 to M26 are converted into digital audio data (hereinafter, the audio data is referred to as original audio data) by the terminals a1 to a26 for analog / digital conversion. 1 is supplied. The terminals a1 to a26 in this example are daisy chain connected to the voice input unit 41 of the voice minutes creating apparatus 1 in the order of microphone numbers. The collected voice data is supplied to the voice minutes creating apparatus 1 via the voice input unit 41 which is a voice input interface. However, the number of microphones is not limited to 26, and the number may be increased or decreased as necessary.

  The audio minutes creating apparatus 1 includes an input unit and an output unit that serve as an interface to an external device on the front surface of the device 1. The input device 4 including a keyboard and a mouse is connected to an input unit 47 that inputs an external signal to the device 1, and supplies the input signal to the device 1. An operation unit 43 having buttons, switches and the like is attached to the front surface of the apparatus 1 and can be directly operated. By the operation input of the input device 4 or the operation unit 43, setting of threshold values of level data, setting of the number of peaks, setting of start / end of voice minutes creation time, and the like described later can be performed. The audio minutes created by the audio minutes creating device 1 are stored in a fourth memory 24 (described later) stored in the device 1. On the monitor 2 equipped with a liquid crystal display panel, the waveform of level data corresponding to all microphone numbers is displayed every hour so that even a speaker who is difficult to hear when the level is below the threshold th is recorded in the audio minutes. Manual operation can be performed. Then, when reproducing the audio minutes, the audio minutes file is read from the fourth memory 24 and emitted from the speaker 3.

  Next, an example of the internal configuration of a terminal that is connected to a microphone and transmits an audio signal will be described with reference to FIG. In this example, the terminals a1 to a3 are connected in a daisy chain, and an example is shown in which data is transmitted and received via the terminal a2 as the terminal a1 as the previous terminal and the terminal a3 as the next terminal. The terminal a2 includes a transmission / reception block 10a that performs data transmission / reception and automatic loopback control at the end, and a data processing block 10B that writes control data to the slot. Further, the terminal a2 has an A terminal 10d for connection with the previous terminal a1 and a B terminal 10e for connection with the next terminal a3. Each terminal is provided with signal lines 11a and 11b for data transmission and a power line 11c. The signal line 11a is for the processing path, and the signal line 11b is for the relay path. The terminal a2 is provided with a power supply terminal 10f for supplying power.

  The terminal a2 includes a terminal 10i for outputting data of the reception slot. In this example, the data master terminal is a terminal arranged on the most downstream side of the transmission line, and is used to output collected data to the outside. On the other hand, the terminal for non-data master is generally a terminal arranged on the upstream side excluding the most downstream of the transmission path. The terminal 10g is a terminal necessary for a non-data master terminal, and the terminal 10i is a terminal necessary for a data master terminal. In this example, input / output of data to the terminals 10g and 10i is detected, and the terminal is switched according to the detection state.

  The terminal a2 includes a terminal 10g for inputting an analog audio signal from a microphone. The analog audio signal collected by the microphone M2 and input from the terminal 10g is converted into a digital audio signal by the analog / digital conversion unit 10c and supplied to the data processing block 10b. Then, the audio signal created by the previous terminal is supplied to the next terminal via the transmission / reception block 10a. At the terminal end (for example, terminal a26), the terminal 10i and the audio input unit 41 of the apparatus 1 are connected to supply an audio signal to the apparatus 1. The supplied audio signal can be stored as original audio data by the audio minutes generating apparatus 1.

  Next, an example of the structure of transmission data used for transmitting data will be described with reference to FIG. In this example, UART (Universal Asynchronous Receiver Transmitter) is applied to the transmission. Since UART itself is a well-known technique as one of asynchronous transmission techniques, detailed description thereof is omitted. Briefly, in this technique, after the start bit “0” is detected, communication can be performed by determining 1/0 in the central phase of the bit by a predetermined number of bits every predetermined time by an internal counter. . After reading the predetermined bits, the detection of the start bit of the next frame is newly started. FIG. 3A shows an example of a frame configuration. In this example, the frame frequency fs is set to 22.05 kHz (4.5 μsec). FIG. 3B shows a data configuration example. One frame is composed of 31 slots and a fixed-length gap (data “1”). Of the 31 slots, 26 slots are audio data, and the remaining 5 slots are control data. The data length of each slot is 17 bits. One slot consists of a 1-bit start bit “0” followed by 16-bit data ds.

  Next, an example of the internal configuration of the audio minutes creating apparatus 1 of this example will be described with reference to FIG. The voice collected by the microphones M1 to M26 of the speech content of the speaker and the surrounding noise is subjected to analog / digital conversion at the terminals a1 to a26, and the voice input unit passes through a cable connecting the terminals a1 to a26 as voice signals. It inputs into the apparatus 1 via 41. The apparatus 1 includes a first memory 21 to a fourth memory 24 that can store a large amount of data in order to store input audio signals and conversion data in various conversion processes of the audio signals. In this example, flash memories are used as the memories 21 to 24, for example.

  A control unit 42 that controls each unit reads a program, a fixed parameter, and the like from a read-only ROM (Read Only Memory) 44 and executes processing, and secures a work area in a writable RAM (Random Access Memory) 45. Then, variables, temporary data, and the like are stored, and the data is read from the RAM 45 and used for processing as necessary. Further, the control unit 42 reads the time from the clock unit 46 that measures the time, and controls the timing of reading and writing to the memories 21 to 24. Then, the control unit 42 of this example creates an original audio data creation unit 42a that writes the original audio data input from the audio input unit 41 to the first memory 21, and creates level data of only the size from the original audio data. A level data creation unit 42b to be written to the second memory 22, an identifier extraction unit 42c to write a certain number of level data microphone numbers exceeding a predetermined threshold value to the peak table created in the third memory 23, and Among the original audio data read from the memory 21 of 1, the original audio data other than the microphone number read from the peak table is attenuated to obtain attenuated audio data, and the attenuated audio data and the original number corresponding to the microphone number read from the peak table are stored. An audio data addition control unit 42d that adds audio data at the same time and creates audio minutes is provided.

  The audio data input to the audio minutes generating apparatus 1 in time order manages the write address of the original audio data generating unit 42a and writes it as original audio data in the memory area corresponding to each microphone number of the first memory 21. It is. In the first memory 21, 26 storage areas from areas m1 to m26 corresponding to the microphone numbers are secured. The original audio data collected by the microphone is written in the area corresponding to each microphone number, in the area of m1, the area of m1, the area of m2, the area of m2, the microphone M26, and the area of m26. Since the audio signal is input from the audio input unit 41 as a digital signal multiplexed on one signal line, it can be used as write data to the first memory 21 in accordance with the multiplexed timing signal.

  The level data creation unit 42b manages the read address of the first memory 21 and reads the original audio data for each microphone number. In the original voice data, the signal of the voice data fluctuates positive and negative, and also includes high frequency noise other than voice. The original voice data is detected into a positive waveform through a low pass filter (LPF) of about 100 Hz to 1 kHz and a detection unit 25 including a rectifier circuit, and level data indicating the magnitude (voice level) is created. . The level data creation unit 42b writes the created level data in an area corresponding to each microphone number in the second memory 22 by managing the write address. In the second memory 22, 26 storage areas from the areas mL1 to mL26 corresponding to the microphone numbers are secured in advance. Then, the microphone M1 is written in the region mL1, the microphone M2 is written in the region mL2,.

  The identifier extraction unit 23 manages the read address of the second memory 22 and reads the level data at the same time and for each microphone number. Then, the microphone numbers whose level data exceeds the threshold are extracted by a predetermined number of peaks, and the extracted microphone numbers are written in the peak table configured in the third memory 23 in time order. In the peak table of this example, the number of peaks is three, that is, p1 to p3, and three microphone numbers are extracted in descending order of the level data. If the size of the level data is less than the threshold value, the microphone number is not extracted. In this way, the microphone numbers that peak at the times t1, t2,..., T are written. In the apparatus 1 of this example, a graph representing the level of all microphones at the same time can be output from the video output unit 31 to the monitor 2 and displayed. Then, the microphone number can be arbitrarily extracted and written to the peak table by an external operation of the operation unit 43 or the input device 4.

  The audio data addition control unit 42d manages the read address of the third memory 23 and reads the microphone numbers written in the peak table in order of time. The audio data addition control unit 42d manages the read address of the first memory 21 and reads the original audio data of all microphones at the same time. Then, the audio data addition control unit 42d performs control for supplying the original audio data corresponding to other than the microphone number stored in the peak table to the attenuator 27 and generating attenuated audio data whose level is reduced from the original audio data. Do. Then, at the same time, the original audio data corresponding to the microphone number stored in the peak table and the attenuated audio data are added by the accumulator 28 to create audio minutes data.

  The created voice minutes data is stored in the fourth memory 24 as a voice minutes file. At the time of voice output, the stored voice minutes file is read as needed, and the digital data is converted into an analog voice signal by the digital / analog converter 29 which converts the digital data into an analog signal. Then, an analog audio signal is supplied to the speaker 3 via the audio output unit 30 that is an interface to the speaker 3, and the speaker 3 emits sound. However, the created audio minutes data can be directly emitted from the speaker 3 via the digital / analog converter 29 and the audio output unit 30 without being stored in the fourth memory 24.

  Next, an example of the peak table creation process of this example will be described with reference to FIGS. FIG. 5 is a flowchart of an example of peak table creation processing. FIG. 6A to FIG. 6C are examples of the waveform of each data, and FIG. 6D is an example of the peak table. First, the original audio data creation unit 42a stores the audio signal input from the audio input unit 41 in the first memory 21 as original audio data (step ST1). At this time, the sound collected by the microphones M1 to M3 and M26 draws a waveform shown in FIG. 6A with the level on the vertical axis and the time on the horizontal axis. Here, the terminals a1 to a26 convert the collected analog audio signal into a digital audio signal with a sampling frequency of, for example, 22.05 kHz and a quantization bit number of 16 bits, and supply the converted signal to the apparatus 1. Then, the original audio data creation unit 42a writes the digitized original audio data in the areas m1 to m26 of the first memory 21 for each microphone number.

  Next, the level data creation unit 42b reads the areas m1 to m26 of the first memory 21 into the microphone numbers (step ST2), rectifies them through the detection unit 26, and the areas mL1 to mL2 of the second memory 22 for each microphone number. Level data is written in mL26 (step ST3).

  By the process of step ST3, the original audio data is converted into level data having an interval of 45 ms and a size of 8 bits. Here, when the sampling frequency is 22.05 kHz, the interval is about 45 μsec. However, by setting the interval to 45 msec, the data amount can be reduced to 1/1000. Furthermore, the data amount can be reduced to ½ by changing the number of quantization bits from 16 bits to 8 bits. For this reason, the amount of level data can be reduced to 1/2000 of the original audio data. At this time, it becomes a positive envelope as in the example of the waveform shown in FIG.

  Next, a certain time t is determined, and the level data of all microphones corresponding to the time t is read from the second memory 22 (step ST4). The identifier extraction unit 42c counts up the time t at regular time intervals if the time t is a variable and the time is within the time 0 <t <T. Then, the level data at the same time t is read from the second memory 22 for all microphone numbers (26 in this example). At a certain time t, first, a variable i is defined as a subscript for reading out the second memory 22 and is set as a region mLi. Then, the initial value 1 is set to i (step ST5), and the level data of the region mL1 at time t is read.

  And it is judged whether it is mLi> mL26 (step ST6). When mLi ≦ mL26, the identifier extraction unit 42c detects a microphone number having a large level value at time t and temporarily stores it in the RAM 45 (step ST7). Here, an example of the level data waveform at each time t1 to t4 shown in FIG. 6C is shown with the vertical axis representing the level and the horizontal axis representing the microphone number. In FIG. 6C, a threshold value th is set in advance, level data that does not exceed the threshold value th is not determined to be a peak, and the microphone number is not written in the peak table.

  However, the example of the waveform of the level data for each time shown in FIG. 6C can be output from the video output unit 31 as a video signal and displayed on the monitor 2. In this case, it is possible to specify that an arbitrary microphone number extracted by the user's manual operation is written in the peak table.

  Then, the subscript i is incremented by one, and the process returns to the determination process of step ST6, and the process is repeated until mLi> mL26. Here, in step ST7, when a level larger than the level of the microphone number temporarily stored in the RAM 45 is detected, the microphone number temporarily stored in the RAM 45 is rewritten with the microphone number. The number of microphone numbers stored in the RAM 45 can be arbitrarily set. In this example, up to three microphone numbers can be temporarily stored in descending order of level.

  When the reading up to the region mL26 and the extraction of the peak microphone number are completed, the microphone number is written in the peak table created in the third memory 23 (step ST8). At this time, among the microphone numbers for each time t temporarily stored in the RAM 45, three microphone numbers are read, and the microphone numbers are written in the peak table as p1, p2, and p3 in descending order of level. However, if the level data does not exceed the threshold value, the microphone number is not written in the peak table.

  Here, an example of the peak table is shown in FIG. The peak table of this example is represented in a table format with peaks p1 to p3 as columns and times t1 to T as rows. At time t1, the microphone M1 is written in the first largest peak p1, and the microphone M26 is written in the second largest peak p2. Similarly, at time t2, microphone M2 is written at peak p1, microphone M3 at peak p2, and microphone M1 is written at the third largest peak p3. At time t3, the microphone M3 is written at the peak p1. At time t4, since there is no level data exceeding the threshold th, nothing is written in the peak table.

  Thus, while counting up the time t, the identifier extraction unit 23c reads the level data until the time t = T and repeats the writing process to the peak table. When the time t = T, the peak table creation process is terminated.

  Next, an example of the audio minutes creation process of this example will be described with reference to the flowchart of FIG. A certain time is determined, and the microphone number of the peak table corresponding to the same time is read from the third memory 23 (step ST11). The voice data addition control unit 42d uses the time t as a variable and counts up the time t at a constant time interval within the time 0 <t <T, and peaks p1 to p3 at the same time t from the peak table. Read the microphone number. At a certain time t, a variable j is defined as a subscript for reading out the first memory 21 and is set as an area mj. Then, the initial value 1 is set to j (step ST13), and the original audio data in the area m1 at time t is read.

  Then, it is determined whether mj> m26 (step ST14). When mj ≦ m26, the voice data addition control unit 42d reads the original voice data corresponding to the microphone number at time t read from the peak table from the first memory 21 (step ST15). Then, the original voice data other than the microphone number described in the peak table is attenuated by the attenuator 27 (step ST16) to obtain attenuated voice data. The attenuation value of the attenuator is determined by the request for reproducing and reproducing a plurality of peaks and ambient noise. The original voice data of the microphone number described in the peak table is not processed.

  Further, the attenuated sound data and the original sound data are added at the same time t to create added sound data (step ST17). Then, the subscript j is incremented by one, and the process returns to the determination process of step ST14 and the process is repeated until mj> m26. Then, when the reading is completed up to the area m26 at time t, it is determined whether to output sound by the speaker 3 or to write to the audio minutes file created in the fourth memory 24 (step ST18).

  When outputting the sound, the added sound data is converted from digital to analog and an analog sound signal is supplied to the speaker 3 to emit sound (step ST19). When writing to the audio file, the added audio data is written to the audio minutes file created in the fourth memory 24 (step ST20).

  Then, until time t = T, the audio data addition control unit 23d sequentially reads the original audio data in the first memory 21 over the entire microphone area, and the selective addition is performed every time t by the accumulator 28. Create data. When the time t = T is reached, the audio minutes creation process is terminated.

  In this way, when the original voice data collected for each speaker is added, the voice minutes can be created by attenuating the voice other than the speaker.

  According to the present embodiment, the audio minutes can be created by adding a plurality of appropriately processed audio data. For this reason, unnecessary noise in the surroundings is suppressed, the speech content is not interrupted, and a voice minutes in which only the necessary speech is recorded can be obtained. In addition, when the sound is collected by a plurality of microphones in a meeting or the like, a voice minutes with a good recording state can be created.

  In addition, since the voice data in which the voice other than the speaker is attenuated among the original voice data read from the first memory 21 and the original voice data of the microphone number written in the peak table are added. The content is more prominently reflected in the audio minutes. Further, for example, if only the maximum peak is conspicuous, the attenuation amount of other signals may be infinite, and if the surrounding background sound is also mixed, it may be non-infinite. Further, when a plurality of peaks cross over time, the attenuation can be continuously changed over time so as not to feel unnaturalness.

  In the above-described embodiment, the connection between the terminals a1 to a26 and the audio minutes creating apparatus 1 is a wired connection, but may be a wireless connection. In this way, there is an effect that the work of installing the cable becomes unnecessary and the installation of the microphone and the terminal becomes easy.

  In the embodiment described above, the flash memory is used as the memory for storing the audio data. However, the audio data may be recorded in a large-capacity recording device such as a hard disk drive or a tape drive.

  In addition, when writing audio minutes data to the audio minutes file created in the fourth memory 24, a plurality of tracks are provided, and the audio minutes file is stored for each track, so that the meeting date, time, etc. are different. Multiple audio minutes files may be created. Further, the microphone minutes and the track numbers may be associated with each other so that the voice minutes of speakers different for each track may be stored. Alternatively, when the peak table is stored, the microphone number that is the peak for each track provided in the fourth memory 24 may be individually stored and used as the peak table.

  In the above-described embodiment, the number of peaks to be extracted set in the peak table is three, but any number of peaks can be set. For example, the identifier extraction unit 42c may write only the microphone number whose level is the maximum value into the peak table, for example, with one peak. When the number of peaks is three, the microphone number whose level is the maximum value and the microphone number of the microphone arranged adjacent to the microphone of this microphone number may be written in the peak table. Alternatively, the number of peaks may be two, and the microphone level having the maximum value and the second largest level may be written in the peak table. As described above, there is an effect that the audio minutes according to the use situation can be created by changing the number of peaks and the writing condition to the peak table. Further, only a specific microphone number may be written in the peak table. In this way, it is also possible to create only the content of the speech of a specific speaker as a voice minutes.

  Further, in the above-described embodiment, the fourth memory 24 is stored as a voice minutes file, but the fourth memory 24 is a stick-type semiconductor memory device such as a stick that can be attached to and detached from the device 1. Also good. In this way, the fourth memory 24 can be arbitrarily removed and loaded in another device to reproduce the audio minutes file. Alternatively, when the fourth memory 24 is attached to the apparatus 1, the original audio data may be automatically read from the first memory 21 to create an audio minutes file in the fourth memory 24.

  In the above-described embodiment, the fourth memory 24 is stored as a voice minutes file. However, the voice contents are automatically read from the voice minutes data output by the voice data addition control unit 42d. A character minutes file may be created. By doing so, there is an effect that the trouble of inputting the contents of the utterance while reproducing the audio minutes file becomes unnecessary.

  In the above-described embodiment, an apparatus for creating a voice minutes, etc. at a meeting or the like is taken as an example. However, any apparatus that processes sound data collected by a plurality of microphones may be used for other purposes. The present invention can also be applied to similar sound processing apparatuses.

It is the block diagram which showed the example of a connection of the audio | voice minutes creation system in one embodiment of this invention. It is the block diagram which showed the example of the internal structure of the terminal in one embodiment of this invention. It is explanatory drawing which showed the structural example of the transmission data in one embodiment of this invention. It is the block diagram which showed the example of an internal structure of the audio | voice minutes production apparatus in one embodiment of this invention. It is the flowchart which showed the creation process example of the peak table in one embodiment of this invention. It is explanatory drawing which showed the creation example of the peak table in one embodiment of this invention. It is the flowchart which showed the example of a production process of the audio minutes in one embodiment of this invention.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Voice minutes creation apparatus, 2 ... Monitor, 3 ... Speaker, 4 ... Input device, 5 ... Conference table, 21-24 ... Memory, 25 ... Detection part, 27 ... Attenuator, 28 ... Accumulator, 29 ... Digital / analog Conversion unit, 30 ... audio output unit, 31 ... video output unit, 29 ... external input unit, 41 ... audio input unit, 42 ... control unit, 42a ... original audio data creation unit, 42b ... level data creation unit, 42c ... identifier Extraction unit, 42d ... audio data addition control unit, 43 ... operation unit, 44 ... ROM, 45 ... RAM, 46 ... clock unit, 47 ... input unit, 100 ... audio minutes creation system, M1-M26 ... microphone, a1- a26 ... terminal

Claims (9)

A first storage unit for storing an original sound data collected by at least two or more microphones with an identifier;
An identifier extraction unit that extracts the identifier assigned to the original audio data in which the level of the original audio data stored in the first storage unit exceeds a preset threshold;
An audio processing apparatus comprising: an audio data addition control unit that attenuates and adds original audio data corresponding to other than the extracted identifier among the original audio data read from the first storage unit . The speech processing apparatus according to claim 1, wherein
A display unit for displaying level data for each same time for each identifier;
An audio processing apparatus comprising: an operation unit that extracts an arbitrary identifier from the level data displayed on the display unit. The speech processing apparatus according to claim 1, wherein
As the identifier extraction unit,
A second storage unit that stores data of the level of the original voice data stored in the first storage unit for each identifier;
A third storage for extracting the identifier of data at a level exceeding the threshold value at each time from the level data for each identifier stored in the second storage unit and storing the extracted identifier And a voice processing device. The speech processing apparatus according to claim 1, wherein
The said identifier extraction part extracts the said identifier from which the said level data becomes the maximum, The audio processing apparatus characterized by the above-mentioned. The speech processing apparatus according to claim 1, wherein
The said identifier extraction part extracts the said identifier with which the said level data becomes the maximum, and the said identifier with the 2nd maximum, The audio processing apparatus characterized by the above-mentioned. The speech processing apparatus according to claim 1, wherein
The identifier extraction unit extracts the identifier having the maximum level data and the identifier assigned to the second microphone arranged in the vicinity of the first microphone to which the selected identifier is assigned. Voice processing device. The speech processing apparatus according to claim 1, wherein
The said identifier extraction part extracts the said specific identifier, The audio processing apparatus characterized by the above-mentioned. An identifier is assigned to the original voice data collected by at least two microphones and stored,
Extracting the identifier given to the original voice data in which the level of the stored original voice data exceeds a preset threshold;
A voice processing method, comprising: attenuating and adding original voice data other than the extracted identifier among the read original voice data. A storage process for storing the original voice data collected by at least two or more microphones with an identifier;
An identifier extraction process for extracting the identifier assigned to the original audio data in which the level of the stored original audio data exceeds a preset threshold;
A program for executing an audio data addition control process for attenuating and adding original audio data corresponding to other than the extracted identifier among the read original audio data.

Images