JPH09106339A - Information processor and data storing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an information processor selecting only the data including the voice signal satisfying a desired condition at the time of an input processing and storing the data and data storing method therefor. SOLUTION: Various kinds of data including a voice signal is inputted by an input part 3 and various kinds of data including this inputted voice signal is held as temporary stored data 8 in a storage part 4. When an analyzing part 5 performs a prescribed analysis for the voice signal to be the temporary stored data 8 that this storage part 4 holds and this analysis result is matched with a preliminarily set condition, various kinds of data including this voice signal is stored in a recording file 11. Thus, data quantity to be stored can be remarkably reduced.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information processing apparatus for fetching and processing various kinds of data such as voices and images and a data storage method applied to the same, and particularly to a desired one from a redundant and huge amount of data. The present invention relates to an information processing apparatus and a data storage method for selecting only data including an audio signal satisfying the condition (3) at the time of input processing and significantly reducing the amount of stored data.

[0002]

2. Description of the Related Art Conventionally, when recording voice or images, as one of them, there is a method in which voice is always recorded like an airplane flight recorder. Due to the restriction, there was a problem that it was not possible to keep a record before a certain period of time. On the other hand, as a method in which the limitation of the recording capacity is taken into consideration, there is a method of recording at a fixed time interval like some security cameras, or recording only when a signal is detected by an infrared sensor. . However, such a method has a problem that the desired recording is not always left.

Further, in the past, when recording a voice signal, it was not possible to select and record the necessary voice even if the voice gender and the utterer were obvious in advance. When recording, there was no choice but to record even unnecessary voice, and the recorded data was redundant and enormous. Further, even when they are searched and reproduced later, since the search key is only the recording time, that is, the time axis, a great deal of labor is required for the search and reproduction. This also applies to images. That is,
Regarding images, the amount of data is much larger than that of voice, and it is highly technical to recognize whether or not to record an image and determine whether or not to record it. I was sorting out the presence or absence.

[0004]

As described above, in the case of recording a voice or an image in the related art, it is necessary to select and record the required voice even if the gender of the voice or the speaker is obvious. However, there was a problem that all data had to be recorded because it could not be done. However, it is obvious that such a method is insufficient as a recording method because the recorded data tends to be redundant.

The present invention has been made in view of the above circumstances, and when recording a voice, the amount of data to be recorded is greatly increased by analyzing the voice and determining whether or not to record the voice. It is also an object of the present invention to provide an information processing apparatus and a data storage method capable of selecting whether or not to record an image by analyzing audio that is simultaneously recorded when recording an image.

[0006]

According to the present invention, there is provided input means for inputting various data including a voice signal, a buffer for temporarily holding various data including the voice signal input by the input means, and the voice. When a secondary storage device that stores various data including signals, an analysis unit that performs a predetermined analysis on the audio signal held in the buffer, and an analysis result of the analysis unit matches a preset condition, Storage means for storing various data including the audio signal in the secondary storage device.

Also, the present invention provides an input means for inputting various data including a voice signal, a buffer for temporarily holding various data including the voice signal input by the input means, and a variety for including the voice signal. When a secondary storage device that stores data, an analysis unit that performs a predetermined analysis on the audio signal held in the buffer, and an analysis result of the analysis unit match a preset condition, from that point of time And a storage unit for storing various data including the audio signal in the secondary storage device for a predetermined period of time.

Further, the present invention is characterized in that the analyzing means has at least one or more of analyzing means such as sex, speaker recognition, voice pitch, loudness, and word recognition. Also, the present invention provides a step of inputting various data including an audio signal, a step of temporarily storing various data including the input audio signal in a buffer, and a step of storing various data including the audio signal in the buffer. In parallel with the storing of the voice signal, a step of analyzing the voice signal already stored in the buffer, and when the analysis result matches a preset condition, various data including the voice signal are stored in the secondary storage device. And a step of storing.

According to the present invention, various data such as voice and images are stored in the buffer as temporary storage data, and the voice signal included in this data is analyzed,
Data that is stored as temporary memory data by tracing back to the time axis when the gender, speaker, pitch and loudness of the speaker's voice, words, etc. are judged and the judgment result matches the preset conditions. To record.

That is, the selection of the recorded data at the time of this input processing makes it possible to greatly reduce the amount of data, and it is possible to easily and efficiently retrieve and reproduce the recorded data required by the operator. Becomes

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a functional configuration of the information processing apparatus according to the embodiment. As shown in FIG. 1, the information processing apparatus 1 of the embodiment includes a control unit 2, an input unit 3, a storage unit 4, an analysis unit 5, a search unit 6, a reproduction unit 7, and an output unit 15. I will do it.

The control unit 2 controls the entire information processing apparatus 1, and the input unit 3 inputs various signals such as voice and images. The storage unit 4 holds various signals input from the input unit 3 as temporary data 8, and stores a personal dictionary file 9 in which information regarding the feature amount of the voice of the speaker is registered, and information regarding the voice feature amount of the word. It manages a registered word dictionary file 10 and a recording file 11 (voice file 12, voice / image file 13, log file 14) that stores signals such as voices and images selected by the device.

The analysis unit 5 analyzes the voice signal, which is the temporary storage data 8 held in the storage unit 4, and determines the gender, the speaker (speaker), the voice feature of the speaker, and the like. The outline of the determination principle in the analysis unit 5 is shown below. (1) Gender Pitch (fundamental frequency of vocal cord vibration) and the like are calculated, voice signals whose pitch is equal to or higher than a predetermined threshold are determined to be female, and other signals are determined to be male. (2) Speaker The speaker is determined by calculating FFT (Fast Fourier Transform), frequency cepstrum, PARCOR coefficient, and the like, and performing matching with the information registered in the personal dictionary file 9. (3) Judgment based on pitch of voice, pitch of voice, and amplitude of voice signal. (4) Speech (word) Word dictionary file 1 using general speech analysis / recognition techniques such as FFT, filter analysis, linear prediction analysis, formants, cepstrum coefficients, and hidden Markov models.
The voice is judged by matching with the information registered in 0.

The search unit 6 uses the recording file 11 managed by the storage unit 4 as a key, with the time, sex, speaker, and the feature amount of the speaker's voice as a key, and the corresponding audio file 12 or voice / image file. Search for 13.
The reproduction unit 7 reproduces the searched audio and image files. Then, the output unit 15 outputs a signal for driving and controlling a recording device externally connected to the information processing device and a signal of a sound or an image itself.

Various signals including a voice signal are sequentially input from the input section 3 and held in the storage section 4 as temporary storage data 8. When the held temporary storage data 8 exceeds a predetermined amount, the oldest temporal data is sequentially erased and replaced with a new input signal. Further, in parallel with this holding operation, the voice signal included in the held input signal is sequentially analyzed by the analysis unit 5, and the feature amount of the voice is extracted. Therefore, the temporarily stored time needs to be longer than the time required for the analysis. For example, as shown in FIG. 2, the sound recorded as a file is recorded by adding sound information as a header to the sound data to be recorded ((a) in FIG. 2), or the sound data is recorded in an existing method (for example, It may be saved as it is as the Windows wave format manufactured and sold by Microsoft Corporation, and the voice information may be recorded as a log file ((b) in FIG. 2).

The time relationship between the audio signal input from the input unit 3 and the recorded audio signal is as shown in FIG. That is, while the analysis unit 5 analyzes the voice input from the input unit 3 and makes various determinations,
The audio signal is always input. Therefore, the storage unit 4 holds the data as temporary storage data 8 for the time required for various determinations by the analysis unit 5, and only when the determination result and the selection condition match, the signal "back in time" is set as the recording file 11. Record. Then, at this time, audio information such as recording time and sex is also recorded in the recording file 11 as a header.

The operation procedure for determining whether or not to record the input audio signal according to sex will be described with reference to FIG. When a voice signal is input from the input unit 3 (step A1), the storage unit 4 holds this voice signal as temporary storage data 8 (step A2). On the other hand, the analysis unit 5
Analyzes the temporary storage data 8 held in the storage unit 4 to determine the sex (step A3, step A4).
If this determination result matches the gender desired by the handler (Y in step A5), a voice information header including the recording time and gender is created (step A6), and the storage unit 4 stores this temporary storage data 8. The audio signal held in the above and the created header are stored in the recording file 11 (step A7). Then, this process is repeated until there is no more temporarily stored data 8 (step A8).

As a result, since only the voice signals of the sex desired by the operator are selected and stored at the time of input processing, the amount of data to be stored can be greatly reduced, and the search and reproduction are efficient. Data processing can be performed.

Next, with reference to FIG. 5, an operation procedure for determining whether or not the input voice signal is to be recorded by the speaker will be described. When a voice signal is input from the input unit 3 (step B1), the storage unit 4 holds this voice signal as temporary storage data 8 (step B2). On the other hand, the analysis unit 5
Analyzes the temporary storage data 8 held in the storage unit 4 (step B3) and extracts the feature amount of the voice data (step B4). Then, with the extracted feature amount,
The information relating to the voice feature amount of each person, which is set as the speaker to be recorded and registered in the personal dictionary file 9 in advance, is compared (step B5).

When the feature amount of the extracted voice data matches that of the speaker desired by the operator (Y in step B6), a voice information header including the recording time and the speaker is created (step S6). B7), this temporarily stored data 8
As a result, the audio signal held in the storage unit 4 and the created header are stored in the recording file 11 (step B8). Whether or not the speaker has finished speaking is detected by the analysis unit 5 analyzing the voice signal input from the input unit 3 by the above-described method. Then, this process is repeated until there is no more temporarily stored data 8 (step B9).

As a result, since only the voice signal of the utterer desired by the operator is selected and stored at the time of input processing, the amount of data to be stored can be greatly reduced as in the case described above, and retrieval and reproduction can be performed. Even in such cases, efficient data processing can be performed.

Next, with reference to FIG. 6, an operation procedure for determining whether or not the input voice signal is to be recorded based on the voice pitch and voice volume will be described. In this case, as the first step, the speaker of the voice signal input by the above-described method is identified (step C5 to step C7). When the speaker of this voice signal matches the speaker desired by the operator (Y in step C8), "speaker match flag".
Set up. This is because there is individual difference in pitch and loudness depending on the speaker. Of course, these steps are unnecessary when determining whether or not to record with a uniform voice pitch and volume regardless of individual differences (N in step C3). Further, it is assumed that these personal data are registered in the personal dictionary file 9 in advance. Voice recording is performed only when this flag is set (when the speaker is designated) and the voice pitch or loudness exceeds (or falls below) the desired threshold, and the voice pitch or loudness falls below the threshold. Until (or until it exceeds) or until the speaker changes (step C9 to step C11). And
This process is repeated until there is no more temporarily stored data 8 (step C12).

As a result, only voice signals having the voice volume and pitch desired by the operator (specific or unspecified vocalist) are selected and stored at the time of input processing, and thus are stored in the same manner as described above. The amount of data can be greatly reduced, and efficient data processing can be performed even in search and reproduction.

Next, with reference to FIG. 7, an operation procedure for determining whether or not to record the input voice signal by the spoken word will be described. In this case, the operator first sets the recording start point and the recording end point (step D1). For example, going back before the word "Dorobo" was recognized, 5
Recording is continued between the minute (start point) and 3 minutes after recognition (end point).

When a voice signal is input from the input unit 3 (step D2), the storage unit 4 holds this voice signal as temporary storage data 8 (step D3). On the other hand, the analysis unit 5 analyzes the temporary storage data 8 held in the storage unit 4 (step D4) and extracts the feature amount of the voice data (step D5). Then, the extracted feature amount is compared with the information about the feature amount of the voice of each word which is set as a word to be recorded and registered in the word dictionary file 10 in advance (step D6).

When the feature amount of the extracted voice data matches the word desired by the operator (Y in step D7), a voice information header including the recording time and the start and end points is created (step D8). ), The audio signal held in the storage unit 4 as the temporary storage data 8 and the created header are stored in the recording file 11 (step D9). Then, this process is repeated until there is no more temporarily stored data 8 (step D10).

As a result, since only the voice signals before and after the word desired by the operator is uttered are selected and stored at the time of the input processing, the amount of data to be stored can be greatly reduced as in the above. It is possible to perform efficient data processing even in search, reproduction, and the like.

Next, referring to FIG. 8, an operation procedure for determining whether or not to record the input voice signal by sex and word will be described. In this case, first, when the voice signal matches the gender desired by the operator according to the determination described with reference to FIG. 4 (Y in steps E5 to E6), the "sex matching flag" is set (step E7). When recording the voice, the flag is set, and the desired word is recognized in the voice signal by the determination described in FIG. 7 (step E
8 to Y in step E10), and only when the "word matching flag" is set (step E11), and is continued until the recording start point and end point range set by the operator or the sex is changed.

At this time, the file has recording time, sex,
Voice information such as the uttered word, the set start point and end point of the recording is recorded together as a header. As a result, the data to be recorded can be further narrowed down, and the amount of data to be stored can be significantly reduced.

Next, referring to FIG. 9, an operation procedure for determining whether or not to record the input voice signal by the speaker and the word will be described. In this case, first, when the voice signal matches that of the speaker desired by the operator according to the determination shown in FIG. 5 (Y in steps F6 to F7),
A "speaker match flag" is set (step F8). Then, when recording the voice, this flag is set, and the desired word in the voice signal is recognized by the determination described in FIG. 7 (Y in step F9 to step F10), and the "word matching flag" is set. Only at this time (step F11), the operation is continued until the recording start point and end point range set by the operator or the speaker changes.

At this time, voice information such as recording time, utterer, uttered word, set start point and end point of recording is recorded together in the file as a header. As a result, similarly to the above, the data can be further narrowed down, and the amount of data to be stored can be significantly reduced.

Next, with reference to FIG. 10, an operation procedure for determining whether or not to record the input voice signal by the voice pitch and voice volume of an arbitrary utterer and a word will be described. In this case, first, the voice signal according to the determination shown in FIG. 5 matches that of the speaker desired by the operator (Y in step G9), and the voice desired by the operator is determined by the determination described in FIG. When the size and height of the voice are recognized (Y in step G10), a "voice pitch / volume flag" is set (step G11). Then, the voice recording is performed only when this flag is set and the desired word is recognized in the voice signal by the determination shown in FIG. 7 and the "word recognition flag" is set (step G16). Continue until the recording start point and end point range set by the operator, the speaker change, or the voice pitch or volume falls below (or exceeds) the threshold.

At this time, in the file, the recording time, the speaker, the uttered word, the set start and end points of the recording, and the voice information regarding the pitch and loudness of the voice are recorded together as a header. As a result, similarly to the above, the data can be further narrowed down, and the amount of data to be stored can be significantly reduced.

As described above, when a voice signal is recorded, the voice signal is immediately analyzed at the time of input processing to determine the sex, the speaker, the pitch of the voice, the size of the voice, the voice (word), etc. By recording only the desired voice, it is possible to significantly reduce the amount of data to be stored, and if the interface shown in FIG. 11 is provided to the operator, the operator can visually recognize the voice. Can be identified,
Only by operating the various buttons 21 to 25, it becomes possible to easily and efficiently retrieve and reproduce the required record data.

In this embodiment, the handling of audio data has been mainly described. For image data including an audio signal, for example, the audio signal is analyzed to determine whether or not the image data itself is recorded. Such an application is possible.

When an image or sound is recorded on an external recording device such as a video camera, the sound or image signal output from the camera is input through the input unit 3 of the same device and recorded according to the procedure described above. Data to be selected, and output the audio / image signal and the signal for operating the external recording device 1
Application such as outputting from 5 is also conceivable. In many cases, an external recording device such as a video device records an audio / image signal by an analog signal, so that the output unit 15 performs digital / analog conversion and outputs the result. That is, by re-inputting only a necessary portion into the video camera via the output unit 15 of the same device, consumption of a recording medium such as a consumed tape can be suppressed.

[0037]

As described in detail above, according to the present invention,
When recording a voice, first, the voice is analyzed, and it is selected whether or not to record the voice by recognizing the gender, the speaker, or the spoken word, and only the selected voice is recorded. As a result, the amount of recorded data can be significantly reduced, and the labor of the operator is reduced during searching and reproduction.

Similarly, for example, when an image is recorded, the voice input at the same time is analyzed to analyze the voice as described above to select whether or not to record. As a result, the image can be recorded without recognizing the image, the amount of data at the time of recording can be significantly reduced, and the labor of the operator at the time of searching and reproducing can be reduced.

When recording voice and images using an external recording device such as a video camera, the image and voice signals are input to this device to analyze the voice, and the voice the operator desires to record. This device outputs a signal after selecting the image and the image, so that the amount of audio / image recorded on the external recording device can be significantly reduced. Therefore, the labor of the operator is reduced when searching and reproducing.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a schematic configuration of an information processing apparatus according to an embodiment of the present invention.

FIG. 2 is a conceptual diagram showing a voice recording method according to the embodiment.

FIG. 3 is a view showing a temporal relationship between an input audio signal and a recorded audio signal according to the same embodiment.

FIG. 4 is a flowchart illustrating an operation procedure when determining whether or not to record an input audio signal according to the embodiment according to sex.

FIG. 5 is a flowchart illustrating an operation procedure when determining whether or not an utterer records an input voice signal according to the embodiment.

FIG. 6 is a flowchart illustrating an operation procedure for determining whether or not to record an input audio signal according to a voice pitch or a voice volume according to the embodiment.

FIG. 7 is an exemplary flowchart illustrating an operation procedure for determining whether to record an input voice signal according to a uttered word according to the embodiment.

FIG. 8 is an exemplary flowchart for explaining an operation procedure when determining whether or not to record the input voice signal according to the embodiment by sex and words.

FIG. 9 is a flowchart illustrating an operation procedure when determining whether to record an input voice signal according to the same embodiment by a speaker and a word.

FIG. 10 is an exemplary flowchart for explaining an operation procedure for determining whether or not to record an input voice signal according to the same embodiment by a voice pitch and a voice volume of an arbitrary speaker and a word.

FIG. 11 is an exemplary view showing an interface provided to the operator by the information processing apparatus of the embodiment.

[Explanation of symbols]

1 ... Information processing device, 2 ... Control part, 3 ... Input part, 4 ... Storage part, 5 ... Analysis part, 6 ... Search part, 7 ... Reproduction part, 8 ... Temporary storage data, 9 ... Personal dictionary file, 10 ... Word dictionary file, 11 ... Recording file, 12 ... Voice file, 13
... audio / image file, 14 ... log file, 15 ... output unit.

Claims (4)

[Claims] 1. An input unit for inputting various data including an audio signal, a buffer for temporarily holding various data including the audio signal input by the input unit, and various data for storing the audio signal. A secondary storage device, an analysis unit that performs a predetermined analysis on the audio signal held in the buffer, and various data including the audio signal when the analysis result of the analysis unit matches a preset condition. An information processing apparatus, comprising: a storage unit that stores the data in the secondary storage device. 2. The storage means stores, when the analysis result of the analysis means matches a preset condition, various data including the audio signal in the secondary storage device by going back a predetermined period from that time. The information processing apparatus according to claim 1, wherein the information processing apparatus stores the information. 3. The information processing apparatus according to claim 1, wherein the analysis unit has at least one of a gender, a speaker recognition, a voice pitch, a voice level, and a word recognition. . 4. A step of inputting various data including an audio signal, a step of temporarily storing various data including the input audio signal in a buffer, and a step of storing various data including the audio signal into the buffer. In parallel with the storage, a step of analyzing the voice signal already stored in the buffer, and when the analysis result matches a preset condition,
Storing various data including the audio signal in a secondary storage device.