JP4779954B2 - Audio data processing apparatus, method and program - Google Patents

Description

  The present invention relates to an audio data processing apparatus, method, and program.

  In recent years, memories and HDDs (Hard Disk Drives) have become large-capacity and inexpensive, and devices such as portable players and voice recorders that digitally compress sound and record for a long time have become widespread. For example, when recording the audio of a conference using a portable player, the long-time conference is recorded as one audio data file, and when listening to the audio later, cueing at the position where you want to hear or at a position with good separation can be made. It becomes difficult. Although it is possible to shorten the cue time by fast-forwarding playback, there is a limit to the playback speed that humans can hear, so excessive fast-forwarding to shorten the time makes accurate cueing difficult. .

In order to solve such a problem, conventionally, a user divides a file by appropriately operating a device at the time of recording audio, but this is troublesome.
Further, Patent Document 1 proposes a method of dividing a file by detecting silence during recording, and a method of dividing a file by detecting a predetermined keyword.
JP 2005-221565 A

However, in the above-described prior art, in the case of a method for detecting a silent section of speech, it is considered that a file is frequently divided by a large number of silent sections when the recorded conversation tends to be interrupted.
In the case of a method for detecting a predetermined keyword, it is necessary to input the predetermined keyword from an operation unit such as a portable player in advance, so that the operation is complicated. In addition, since words frequently used at the delimiter position differ depending on the speaker's habit, it is preferable that keyword setting can be performed by determining the speaker's habit during recording.

  An object of the present invention is to easily and appropriately classify audio data.

The invention described in claim 1
In an audio data processing apparatus having an input means and an audio data recording means for recording audio data input by the input means,
Voice recognition means for performing voice recognition of the voice data input by the input means;
Operation means for accepting an operation at the time of input by the input means;
Voice recognition data registration means for registering voice recognition data recognized by the voice recognition means based on an operation at the time of input by the operation means;
Based on the voice recognition result of the input voice data in the voice recognition means and the voice recognition data registered in the voice recognition data registration means, the input voice data is divided and recorded by the recording means. Recording control means for controlling so as to
It is characterized by providing.

The invention described in claim 2
In an audio data processing apparatus having reproduction means for reproducing audio data input from the outside or audio data recorded in advance in the recording means,
Voice recognition means for performing voice recognition of the voice data reproduced by the reproduction means;
Operation means for accepting an operation at the time of sound reproduction by the reproduction means;
Voice recognition data registration means for registering voice recognition data recognized by the voice recognition means based on an operation at the time of voice reproduction by the operation means;
Based on the voice recognition result of the reproduced voice data in the voice recognition means and the voice recognition data registered by the voice recognition data registration means, the voice data is divided and recorded by the recording means. Recording control means for controlling;
It is characterized by providing.

The invention according to claim 3 is the audio data processing device according to claim 1 or 2,
It has a buffer for recording audio data,
The buffer records audio data for a predetermined time according to an instruction from the operation means,
The voice recognition means performs voice recognition on the data recorded in the buffer,
The voice recognition data registration unit registers the voice recognition data recognized by the voice recognition unit.

According to a fourth aspect of the present invention, in the audio data processing device according to any one of the first to third aspects,
The audio data classification performed by the recording control means is file division of audio data.

The invention according to claim 5 is the audio data processing device according to any one of claims 1 to 3,
The audio data classification performed by the recording control means is track rewriting of audio data.

The invention according to claim 6 is the audio data processing device according to any one of claims 1 to 3,
The audio data classification performed by the recording control means is a flag setting for the audio data.

The invention described in claim 7
In an audio data processing method including an input process and an audio data recording process for recording audio data input in the input process,
A voice recognition step of performing voice recognition of the voice data input in the input step;
An operation process for receiving an operation at the time of input in the input process;
A voice recognition data registration step of registering voice recognition data recognized by the voice recognition means based on an operation at the time of input in the operation step;
Based on the voice recognition result of the input voice data in the voice recognition step and the voice recognition data registered in the voice recognition data registration step, the input voice data is divided and recorded in the recording step. A recording control process for controlling
It is characterized by providing.

The invention according to claim 8 provides:
In an audio data processing method having a reproduction step of reproducing audio data input from the outside or audio data recorded in advance in a recording step,
A voice recognition step of performing voice recognition of the voice data reproduced in the reproduction step;
An operation step of accepting an operation at the time of audio reproduction in the reproduction step;
A voice recognition data registration step of registering voice recognition data recognized in the voice recognition step based on an operation at the time of voice reproduction by the operation step;
Based on the voice recognition result of the reproduced voice data in the voice recognition step and the voice recognition data registered in the voice recognition data registration step, the voice data is divided and recorded in the recording step. Recording control process to control;
It is characterized by providing.

The invention according to claim 9 is a program that causes a computer to function as an input unit and an audio data recording unit that records audio data input by the input unit.
Voice recognition means for performing voice recognition of the voice data input by the input means;
Operation means for accepting an operation at the time of input by the input means;
Voice recognition data registration means for registering voice recognition data recognized by the voice recognition means based on an operation at the time of input by the operation means;
Based on the voice recognition result of the input voice data in the voice recognition means and the voice recognition data registered in the voice recognition data registration means, the input voice data is divided and recorded by the recording means. Recording control means for controlling to
It is made to function as.

According to a tenth aspect of the present invention, there is provided a program for causing a computer to function as reproduction means for reproducing audio data input from the outside or audio data recorded in advance in the recording means.
Voice recognition means for performing voice recognition of the voice data reproduced by the reproduction means;
Operation means for accepting an operation at the time of sound reproduction by the reproduction means;
Voice recognition data registration means for registering voice recognition data that has been voice-recognized by the voice recognition means based on an operation during voice reproduction by the operation means;
Based on the voice recognition result of the reproduced voice data in the voice recognition means and the voice recognition data registered by the voice recognition data registration means, the voice data is divided and recorded by the recording means. Recording control means for controlling,
It is made to function as.

  According to the present invention, there is an effect that audio data can be classified easily and appropriately.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. However, the scope of the invention is not limited to this embodiment.

With reference to FIG. 1, the internal configuration of the audio data processing apparatus will be described. The audio data processing apparatus is an apparatus capable of recording sound, inputting and reproducing audio data, such as a portable player, a voice recorder, and a stereo apparatus.
As shown in FIG. 1, an audio data processing device 1 includes an analog input unit 2, an ADC (Analog Digital Converter) 3, an encoding unit 4, a recording unit 5, a digital input unit 6, a word DB (Database) 7, a voice recognition unit. 8, a playback unit 9, a DAC (Digital Analog Converter) 10, an amplification unit 11, an output unit 12, a key operation unit 13, a buffer 14, a control unit 15, and the like.

The analog input unit 2 is a device such as a microphone, and performs processing for collecting external sound and recording it into analog sound data when recording sound.
The ADC 3 performs processing for converting the analog voice data received from the analog input unit 2 into digital voice data.
The encoding unit 4 encodes the digital audio data received from the buffer 14 into a file format such as MP3 (MPEG Audio Layer-3), WMA (Windows Media (registered trademark) Audio), or WAVE, and records the file as a file. Do.
The recording unit 5 is an HDD (Hard Disk Drive) or memory for recording the audio data file generated by the encoding unit 4, the audio data file input from the digital input unit 6, etc. Disk) and other devices that store media.
The digital input unit 6 is an interface such as a USB (Universal Serial Bus), is connected to an external device, and receives digital audio data in a format such as a file.

The reproduction unit 9 performs processing for reproducing audio data of various formats such as MP3, WMA, and WAVE stored in the recording unit 5 under the control of the control unit 15.
The DAC 10 performs processing for converting digital audio data input from the reproduction unit 9 into analog audio data.
The amplifying unit 11 is a device such as an amplifier, and is a device that amplifies analog audio data input from the DAC 10.
The output unit 12 is a device such as a speaker, and is a device that outputs analog audio data input from the amplification unit 11 to the outside as audio.

The word DB 7 is a database in which a plurality of words and their readings are registered, and the words can be searched from the word readings. In addition, keywords to be described later are registered, and keywords can be searched from the readings.
The speech recognition unit 8 performs speech recognition processing on the speech data input from the ADC 3 or the buffer 14 with reference to the word DB 7, so that the likelihood (score) of words that may be represented by the speech data. Is identified. A word is specified by removing a word having a score value lower than a predetermined value from candidates. The speech recognition method is arbitrary, and a word only needs to be specified.

  The key operation unit 13 is a device such as a button or a switch for the user to operate the audio data processing device 1.

The buffer 14 is a storage medium such as a RAM (Random Access Memory) that temporarily stores the digital audio data output from the ADC 3 or the playback unit 9.
The control unit 15 includes a CPU (Central Processing Unit), a ROM (Read Only Memory), a buffer 14, and the like. In accordance with a control program stored in the ROM, the process of dividing the audio data stored in the recording unit 5 by the buffer 14 and the entire process performed by the audio data processing apparatus 1 are controlled.

  Referring to FIG. 2, when voice data processing apparatus 1 is recording, a keyword is detected from the recorded voice, and the voice is divided at the position where the keyword is detected, and saved as a file. The flow of processing when doing this will be described. Note that the control unit 15 executes the process.

This process is started by receiving a recording start instruction from the key operation unit 13.
As shown in FIG. 2, first, a file of a format such as MP3, WMA, or WAVE is generated in the storage unit 6, and the file is opened (S1).

  Next, analog audio data starts to flow continuously from the analog input unit 2, is converted into digital audio data by the ADC 3, and is temporarily stored in the buffer 14 (for example, for 2 seconds). Thereafter, the digital audio data is encoded by the encoding unit 4 and sequentially stored in a file (S2).

Next, when a keyword registration instruction is not received from the key operation unit 13 (S3; No), the process proceeds to S9.
When a keyword registration instruction is received (S3; Yes), the voice data temporarily stored in the buffer 14 is subjected to voice recognition processing by the voice recognition unit 8, and the word represented by the voice data is specified. (S4).
Next, the word specified in S4 is newly registered as a keyword in the word DB 7 (S5).

Next, the accumulation of the voice data in the file is stopped at the position immediately before the word specified in S4, and the file is closed there (S6).
Next, a new file is opened (S7), and the audio data temporarily stored in the buffer 14 is encoded by the encoding unit 4 from the position where the storage was stopped in S6, and sequentially stored in this file ( S8).

Next, the voice data converted into digital data by the ADC 3 is continuously input to the voice recognition unit 8 in parallel with the input to the buffer 14 in parallel with the voice data, and voice recognition processing is performed. When the keyword registered in S5 is not detected (S9; No), the process proceeds to S13.
If a keyword is detected (S9, Yes), the storage of the audio data stored in the buffer 14 at the position immediately before the detected keyword is stopped and the file is closed there (S10).

  Next, a new file is opened (S11), and the audio data temporarily stored in the buffer 14 is encoded by the encoding unit 4 from the position where the storage was stopped in S10, and sequentially stored in this file ( S12).

Next, when there is no instruction to end recording from the key operation unit 13 (S13; No), the process returns to S3.
If an instruction to end recording is received (S13; Yes), the file is closed (S14), and this process ends.

  As described above, if this processing is required, keyword registration processing (S4, S5) is performed when a keyword registration instruction is received, and voice data is divided when a registered keyword is detected by voice recognition. The process of recording as a file (S1, S2, S6, S7, S8, S10, S11, S12, S14) will be performed.

  With reference to FIG. 3, for example, when recording the contents of a conference during a conference, a method for using the voice recorder that is the voice data processing device 1 and an outline of the processing performed by the voice recorder will be described.

First, FIG. 3A is a diagram showing how a keyword is registered from a speaker's utterance and the keyword is detected.
The speaker utters "I think ... Then it will be ... Next ... Now then ..." (a1).
When the person who operates the voice recorder operates the key operation unit 13 in order to register a keyword at the timing of “then” when the speaker speaks “then next” (a2), the buffer 14 is temporarily stored. A voice recognition process is performed by the voice recognition unit 8 on the contents of the voice data (a3) stored (for example, for 2 seconds), and the word “Now” is specified, and the voice in which the word is detected is used. It is registered as a keyword in the word DB 7. Then, the voice recognition unit 8 detects “Now” from the subsequent speech of the speaker (a4).

FIG. 3B is an image diagram showing that a file is recorded every time a keyword is detected from a speaker's utterance.
Each time the keyword “Now” is detected from the utterance of the speaker, the audio data encoded by the encoding unit 4 is divided at that position by the control unit 15 and recorded in the recording unit 5 as a file in the MP3 format ( For example, the file name is “file N.mp3”).

  In the present embodiment, the method of dividing the audio data was a method of recording the divided audio data divided into files, but is not limited to this method, a method of setting flag information in the audio data, It may be a method of recording music data on an MD divided into tracks.

  When the voice data is classified by setting the flag, the process performed by the control unit 15 performs a process of setting flag information in the voice data instead of dividing the voice data. Then, flag information setting is performed instead of steps S6 and S7 in the flowchart of FIG. 2, and similarly, flag information setting is performed instead of steps S10 and S11.

  FIG. 3C is an image diagram showing that audio data is divided by setting flag information. Each time the keyword “Now” is detected from the speaker's utterance, the control unit 15 sets the flag information (c1, c2) at the position of the voice data encoded by the encoding unit 4 and is recorded in the recording unit 5. The

  Also, when recording audio data on an MD divided into tracks, the encoding unit 4 performs an encoding process for recording the audio data on the MD, and the recording unit 5 serves as a device for storing the MD such as an MD drive. The unit 15 performs a process of rewriting address information and track information of audio data held by the MD. Then, in place of steps S6 and S7 in the flowchart of FIG. 2, the address information and track information of the audio data are rewritten, and similarly, the address information and track information of the audio data are rewritten in place of steps S10 and S11. Is done.

  In the present embodiment, the audio data recorded in the recording unit 5 is the audio data input from the analog input unit 2, but is not limited to this method.

  For example, an audio file such as MP3 input from the digital input unit 6 and recorded in the recording unit 5 is reproduced by the reproducing unit 9, and the audio is recognized by the audio recognizing unit 8 while being divided by the encoding unit 4, and the recording unit 5 may be recorded. At this time, the user of the audio data processing apparatus 1 performs keyword registration while listening to the audio reproduced by the reproduction unit 9, converted into analog data by the DAC 10, amplified by the amplification unit 11, and output from the output unit 12. The key operation unit 13 is operated.

  Referring to FIG. 4, when audio data processing apparatus 1 is reproducing audio data, a keyword is detected from the reproduced audio, the audio is divided at the position where the keyword is detected, and a file is The flow of processing when saving as will be described. Note that the control unit 15 executes the process.

This process is started by receiving an instruction to start recording from the key operation unit 13 during reproduction of the audio data recorded in the recording unit 5.
As shown in FIG. 4, first, a file of a format such as MP3, WMA, or WAVE is generated in the storage unit 6, and the file is opened (S21).

  Next, the audio data reproduced by the reproducing unit 9 is temporarily stored in the buffer 14 (for example, for 2 seconds), then encoded by the encoding unit 4, and sequentially stored in the opened file (S22).

Next, when a keyword registration instruction is not received from the key operation unit 13 (S23; No), the process proceeds to S29.
When a keyword registration instruction is received (S23; Yes), the voice data temporarily stored in the buffer 14 is subjected to voice recognition processing by the voice recognition unit 8, and the word represented by the voice data is specified. (S24).

  Next, the word specified in S24 is newly registered as a keyword in the word DB 7 (S25).

  Next, the accumulation of the voice data in the file is stopped at the position immediately before the word specified in S24, and the file is closed there (S26).

  Next, a new file is opened (S27), and the audio data temporarily stored in the buffer 14 is encoded by the encoding unit 4 from the position where the storage was stopped in S26, and sequentially stored in this file ( S28).

Next, the audio data reproduced by the reproducing unit 9 is continuously input in parallel to the audio recognizing unit 8 while being synchronized with the input to the buffer 14, and audio recognition processing is performed. When the keyword registered in S25 is not detected (S29; No), the process proceeds to S33.
When a keyword is detected (S29, Yes), the accumulation of the audio data stored in the buffer 14 at the position immediately before the detected keyword is stopped and the file is closed there (S30).

  Next, a new file is opened (S31), and the audio data temporarily stored in the buffer 14 is encoded by the encoding unit 4 from the position where the storage was stopped in S28, and sequentially stored in this file ( S32).

Next, when there is no instruction to end reproduction from the key operation unit 13 (S33; No), the process returns to S23.
If an instruction to end reproduction is received (S33; Yes), the file is closed (S34), and this process ends.

  The audio data reproduced by the reproduction unit 9 may be audio data input from the analog input unit 2 and recorded in the recording unit 5. The audio data input from the digital input unit 6 may be streaming data instead of a file.

  Also, not a single keyword but a plurality of different keywords may be registered, and the voice data may be recorded by changing the division position of the voice data for each keyword.

  As described above, according to the present embodiment, the audio data input from the analog input unit 2 is encoded by the encoding unit 4, and the audio data is recognized by the audio recognition unit 8 while being recorded in the recording unit 5. By performing an operation for registering a keyword with the operation unit 13, it is possible to register a keyword while recording a voice, and a prior work for registering a keyword for classifying voice data before recording becomes unnecessary.

  Also, by registering keywords while listening to the sound being recorded or played back, what words the speaker uses frequently or what words the speaker uses when changing topics, It is possible to register keywords by judging the speaker's habits.

  Further, by storing the audio data after being converted into digital audio data by the ADC 3 or after being reproduced by the reproduction unit 9 in the buffer 14, for the audio already recorded or reproduced before a certain time, The voice recognition unit 8 can easily perform voice recognition for keyword registration.

  Furthermore, by recording multiple different keywords and recording the voice data by changing the voice data segmentation position for each registered keyword, it is easy to classify the voice at various positions according to the purpose of use of the voice data. It becomes.

It is a block diagram of an audio | voice data processing apparatus. It is a flowchart of the process which records the audio | voice data in recording. (A) is the figure which showed how a keyword was registered from a speaker's utterance and a keyword was detected. (B) is an image diagram showing that a file is recorded every time a keyword is detected from a speaker's utterance. (C) is an image diagram showing that audio data is divided by flag setting. It is a flowchart of the process which records the audio | voice data in reproduction | regeneration.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Audio | voice data processing apparatus 2 Analog input part 3 ADC
4 Encoding unit 5 Recording unit 6 Digital input unit 7 Word DB
8 Voice recognition unit 9 Playback unit 10 DAC
DESCRIPTION OF SYMBOLS 11 Amplification part 12 Output part 13 Key operation part 14 Buffer 15 Control part

Claims (10)

In an audio data processing apparatus having an input means and an audio data recording means for recording audio data input by the input means,
Voice recognition means for performing voice recognition of the voice data input by the input means;
Operation means for accepting an operation at the time of input by the input means;
Voice recognition data registration means for registering voice recognition data recognized by the voice recognition means based on an operation at the time of input by the operation means;
Based on the voice recognition result of the input voice data in the voice recognition means and the voice recognition data registered in the voice recognition data registration means, the input voice data is divided and recorded by the recording means. Recording control means for controlling so as to
An audio data processing apparatus comprising: In an audio data processing apparatus having reproduction means for reproducing audio data input from the outside or audio data recorded in advance in the recording means,
Voice recognition means for performing voice recognition of the voice data reproduced by the reproduction means;
Operation means for accepting an operation at the time of sound reproduction by the reproduction means;
Voice recognition data registration means for registering voice recognition data recognized by the voice recognition means based on an operation at the time of voice reproduction by the operation means;
Based on the voice recognition result of the reproduced voice data in the voice recognition means and the voice recognition data registered by the voice recognition data registration means, the voice data is divided and recorded by the recording means. Recording control means for controlling;
An audio data processing apparatus comprising: It has a buffer for recording audio data,
The buffer records audio data for a predetermined time according to an instruction from the operation means,
The voice recognition means performs voice recognition on the data recorded in the buffer,
The voice data processing apparatus according to claim 1, wherein the voice recognition data registration unit registers voice recognition data voice-recognized by the voice recognition unit.   4. The audio data processing apparatus according to claim 1, wherein the audio data classification performed by the recording control means is file division of audio data.   4. The audio data processing apparatus according to claim 1, wherein the audio data classification performed by the recording control means is track switching of audio data.   4. The audio data processing apparatus according to claim 1, wherein the classification of the audio data performed by the recording control means is a flag setting for the audio data. In an audio data processing method including an input process and an audio data recording process for recording audio data input in the input process,
A voice recognition step of performing voice recognition of the voice data input in the input step;
An operation process for receiving an operation at the time of input in the input process;
A voice recognition data registration step of registering voice recognition data recognized by the voice recognition means based on an operation at the time of input in the operation step;
Based on the voice recognition result of the input voice data in the voice recognition step and the voice recognition data registered in the voice recognition data registration step, the input voice data is divided and recorded in the recording step. A recording control process for controlling
An audio data processing method comprising: In an audio data processing method having a reproduction step of reproducing audio data input from the outside or audio data recorded in advance in a recording step,
A voice recognition step of performing voice recognition of the voice data reproduced in the reproduction step;
An operation step of accepting an operation at the time of audio reproduction in the reproduction step;
A voice recognition data registration step of registering voice recognition data recognized in the voice recognition step based on an operation at the time of voice reproduction by the operation step;
Based on the voice recognition result of the reproduced voice data in the voice recognition step and the voice recognition data registered in the voice recognition data registration step, the voice data is divided and recorded in the recording step. Recording control process to control;
An audio data processing method comprising: In a program for causing a computer to function as input means and sound data recording means for recording sound data input by the input means,
Voice recognition means for performing voice recognition of the voice data input by the input means;
Operation means for accepting an operation at the time of input by the input means;
Voice recognition data registration means for registering voice recognition data recognized by the voice recognition means based on an operation at the time of input by the operation means;
Based on the voice recognition result of the input voice data in the voice recognition means and the voice recognition data registered in the voice recognition data registration means, the input voice data is divided and recorded by the recording means. Recording control means for controlling to
A program characterized by functioning as In a program for causing a computer to function as reproduction means for reproducing audio data input from outside or audio data recorded in advance in a recording means,
Voice recognition means for performing voice recognition of the voice data reproduced by the reproduction means;
Operation means for accepting an operation at the time of sound reproduction by the reproduction means;
Voice recognition data registration means for registering voice recognition data that has been voice-recognized by the voice recognition means based on an operation during voice reproduction by the operation means;
Based on the voice recognition result of the reproduced voice data in the voice recognition means and the voice recognition data registered by the voice recognition data registration means, the voice data is divided and recorded by the recording means. Recording control means for controlling,
A program characterized by functioning as

Images