JP2001222300A - Voice reproducing device and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reproduce a voice at high speed in a time frame nearly equal to the time frame required for reproducing a whole recorded voice at a specified speed while making the voice easy to listen by reducing omission of information as much as possible when the recorded voice is reproduced at the specified speed. SOLUTION: When reproducing instruction of the voice data becoming an n-fold speed (specified speed) reproducing object is inputted, the voice data supplied through a communication line, a CD-RW and a DVD, etc., are separated to a voice section from a non-voice section by a voice analytic part 2a. Then, the speech speed of voice section held between non-voice sections of a fixed time length or above is converted so that the speed becomes later at its top part than the prescribed reproducing speed, and is returned gradually to the prescribed reproducing speed toward the end by a speech speed conversion part 3a. The speech speed converted voice signal and the voice signal of the non-voice section separated by the voice analytic part 2a are synthesized by a synthetic part 5a, and then, the voice is reproduced at the n-fold speed without delaying remarkably from the reproducing time frame while making an important part easy to listen.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a sound reproducing apparatus and a recording medium for assisting a user to hear a high-speed sound which greatly exceeds a normal utterance speed.

[Summary of the Invention] The present invention records an audio signal recorded in advance by an arbitrary signal processing method without changing the pitch of the voice or intentionally changing the pitch of the voice. When playing back faster than usual, the beginning of an utterance and the audio part that is considered important for hearing are automatically detected from information such as the pitch and strength of the voice, and that part is uttered before and after. While converting slower than the speed, non-voice parts and low power parts at the end of the speech are adaptively deleted, so that the contents of high-speed voices up to about 10 times high-speed playback voices, which were previously impossible to hear, can be grasped. Is what makes it possible.

[0003]

2. Description of the Related Art Conventionally, a method of listening to audio at a high speed is a method of playing back an audio recorded on an analog tape recorder or VTR at a high speed, and listening to a voice whose voice pitch has changed in proportion to the playback speed. In general, a method of listening to discrete sounds by skipping data recorded in a digital tape recorder or VTR in proportion to the reproduction speed has been used.

[0004] A speech speed conversion method (Japanese Patent No. 2955247) for successively extracting fundamental frequencies and performing thinning-out processing on a waveform basis to reproduce continuous high-speed voice without changing the pitch of voice. ) Has been proposed.

[0005]

Among the fast listening methods of such a sound, the method using an analog tape recorder has no data loss, but is very hard to hear due to a change in pitch. In the system using the digital tape recorder, information loss increases as the reproduction speed increases, so that both systems have a limit of listening up to about double speed.

[0006] In the above-mentioned speech speed conversion method, it is expected that a higher-speed voice can be heard.
When performing a content search such as TR, the search mainly for video is performed at about 10 times speed, while the main body of audio is at most,
Only a search up to about 2x speed could be performed.

[0007] In addition, there has been proposed a method of extending a voice section by partially deleting information (1).
995 IEEJ D-695). In this method, high-speed reproduction of voice is performed by controlling the voice speed according to the amount of voice data stored in a memory or deleting a silent section.

However, in this method, information is often deleted or the speech speed is controlled irrespective of the importance of the content, so that the content cannot be heard or grasped sufficiently.

As is apparent from the above, conventionally,
In any of the proposed systems, high-speed reproduction of sound can be performed only at a reproduction speed of about 2 × or 3 × which can make the contents understandable.

[0010] For this reason, there has been a strong demand for the development of an apparatus capable of retrieving audio contents at a reproduction speed of about 10 times the same as that for retrieving video contents.

[0011] In view of the above circumstances, the present invention minimizes the loss of information and minimizes the time required to reproduce the entire recorded voice at the specified speed while making it easier to hear the voice reproduced at the specified speed. It is an object of the present invention to provide an audio reproducing apparatus and a recording medium which can be played back at high speed in substantially the same time frame as the frame, thereby enabling a program editor or the like to search the contents of the recorded audio at high speed.

[0012]

In order to achieve the above object, according to the present invention, an audio signal to be reproduced is acoustically analyzed, and a voice signal in a voice section and a voice signal in a non-voice section are provided. And an audio analysis unit that separates the audio signal into audio sections separated by the audio analysis unit. The head of the audio signal is slower than a predetermined reproduction speed, and gradually returns to the predetermined reproduction speed toward the end. A speech speed conversion unit for converting speech speed, and a speech signal converted by speech speed conversion by the speech speed conversion unit and a speech signal in a non-speech section separated by the acoustic analysis unit. And a synthesizing unit for generating a signal.

According to a second aspect of the present invention, in the audio reproducing apparatus according to the first aspect, the non-speech separated by the acoustic analysis unit based on time information of the audio signal subjected to the speech speed conversion by the speech speed conversion unit. A non-speech section length control unit is provided that adaptively deletes or compresses the speech signal in the section and combines the speech signal with the speech signal after the speech speed conversion in the combining unit.

According to a third aspect of the present invention, the power value of the audio signal to be reproduced is determined, and a power threshold value required to separate the audio signal into an audio signal in an audio section and an audio signal in a non-audio section is determined. Using the voice / non-voice determination unit to generate and the power threshold obtained by the voice / non-voice determination unit, the voice signal to be reproduced is converted into a voice signal in a voice section and a voice signal in a non-voice section. A sound analysis unit to be separated, and the head of the sound signal of the sound section separated by the sound analysis unit is slower than a predetermined reproduction speed, and gradually returns to a predetermined reproduction speed toward the end. A speech speed conversion unit for converting a speech speed, a speech signal subjected to speech speed conversion by the speech speed conversion unit, and a speech signal in a non-speech section separated by the acoustic analysis unit; And a synthesizing unit for generating It is characterized.

According to a fourth aspect of the present invention, in the audio reproducing apparatus according to the third aspect, the audio signal to be reproduced is separated into an audio signal in an audio section and an audio signal in a non-audio section by the audio / non-audio section. When generating the power threshold necessary for causing the power threshold to adaptively increase the power threshold in proportion to the amount of delay time accumulation from the original voice due to the expansion of the voice signal in the voice section. It is characterized by changing.

According to a fifth aspect of the present invention, a sound analysis unit that performs sound analysis of a sound signal to be reproduced and separates the sound signal into a sound signal in a sound section and a sound signal in a non-speech section, and a sound separated by the sound analysis unit. A speech speed conversion unit that converts the speech speed so that the beginning portion of the audio signal in the section is slower than a predetermined reproduction speed, and gradually returns to the predetermined reproduction speed toward the end.
Based on the time information of the voice signal subjected to voice speed conversion by the voice speed conversion unit, when adaptively deleting or compressing the voice signal of the non-voice section separated by the acoustic analysis unit,
A non-speech section adjusting unit that outputs the speech signal without being shorter than a predetermined length, a speech signal that has undergone speech rate conversion by the speech rate conversion unit, and a speech signal in a non-speech section obtained by the non-speech section adjustment unit And a synthesizing unit for synthesizing and generating a converted audio signal.

According to a sixth aspect of the present invention, a sound analysis unit that performs sound analysis of a sound signal to be reproduced and separates the sound signal into a sound signal in a sound section and a sound signal in a non-speech section, and a sound separated by the sound analysis unit. A fundamental frequency calculation unit for calculating a fundamental frequency included in the sound signal of the section, and a sound signal of the sound section separated by the sound analysis unit according to a change rate of the fundamental frequency obtained by the fundamental frequency calculation unit. A speech speed conversion unit that adaptively expands the speech speed to convert the speech speed, a speech signal subjected to speech speed conversion by the speech speed conversion unit, and a speech signal of a non-speech section separated by the acoustic analysis unit. A synthesizing unit for synthesizing and generating a converted audio signal.

According to a seventh aspect of the present invention, in the audio reproducing apparatus according to the sixth aspect, when the speech speed converting section expands the audio signal in the audio section separated by the acoustic analysis section, the audio signal is obtained by the fundamental frequency calculating section. The calculated change rate of the fundamental frequency is compared with a preset change rate threshold value, and the change rate of the fundamental frequency obtained by the fundamental frequency calculation unit exceeds the preset change rate threshold value. It is characterized in that the expansion rate for the audio signal in the section is larger than the expansion rate for the preceding and following audio signals.

According to an eighth aspect of the present invention, in the audio reproducing apparatus according to the sixth aspect, when the speech speed converting section expands the audio signal in the audio section separated by the acoustic analysis section, the expansion is performed by a fundamental frequency calculating section. The calculated change rate of the fundamental frequency is compared with a preset change rate threshold value, and the change rate of the fundamental frequency obtained by the fundamental frequency calculation unit exceeds the preset change rate threshold value. In this case, the audio signal is expanded at the same expansion rate for a certain period of time from the appearance time of the voice section or until a certain number of voiced sound sections appear after the voice section.

According to a ninth aspect of the present invention, in a recording medium in which a program for operating a computer device is stored, when a sound reproduction instruction is input and installed in the computer device, a sound to be reproduced is stored in the computer device. A sound analysis unit that performs sound analysis on a signal and separates the sound signal into a speech signal in a speech section and a speech signal in a non-speech section. A speech speed conversion unit that changes the speech speed so as to be slower than the playback speed and gradually returns to a predetermined playback speed toward the end; an audio signal whose speech speed has been converted by the speech speed conversion unit; An audio reproduction program for synthesizing the audio signal of the non-speech section separated by the analysis section and generating a synthesis section for generating a converted audio signal is stored. .

According to a tenth aspect of the present invention, in the recording medium according to the ninth aspect, the voice reproduction program is installed in the computer device, and when a voice reproduction instruction is input, the voice speed is stored in the computer device. Based on the time information of the voice signal subjected to the voice speed conversion by the conversion unit, the voice signal of the non-voice section separated by the sound analysis unit is adaptively deleted or compressed, and the voice speed conversion is performed by the synthesis unit. It is characterized in that a non-voice section length control unit to be synthesized with the subsequent voice signal is generated.

According to an eleventh aspect of the present invention, in a recording medium storing a program for operating a computer device, when a sound reproduction instruction is input and installed in the computer device, a sound to be reproduced is stored in the computer device. An audio / non-speech determining unit that determines a power value of a signal and generates a power threshold necessary for separating the audio signal into an audio signal in an audio section and an audio signal in a non-audio section; Using the power threshold value obtained by the non-speech determination unit, an audio analysis unit that separates the audio signal to be reproduced into an audio signal in a voice section and an audio signal in a non-voice section, and the sound analysis unit The speech speed is converted so that the beginning of the speech signal in the speech section is slower than the predetermined playback speed, and gradually returns to the predetermined playback speed toward the end. A speech speed conversion unit, and a synthesis unit that synthesizes the speech signal subjected to speech speed conversion by the speech speed conversion unit and the speech signal of the non-speech section separated by the sound analysis unit to generate a converted speech signal. And a sound reproducing program for generating the following is stored.

According to a twelfth aspect of the present invention, in the recording medium according to the eleventh aspect, the audio reproduction program is installed in the computer device, and when an audio reproduction instruction is input, the audio reproduction program is stored in the computer device as a reproduction target. Generate a power threshold necessary to separate the audio signal into a voice signal in a voice section and a voice signal in a non-voice section,
Generating the voice / non-voice determination unit that adaptively changes the power threshold in proportion to the amount of delay time storage from the original voice due to the expansion of the voice signal in the voice section. Features.

According to a thirteenth aspect of the present invention, in a recording medium storing a program for operating a computer device, when a sound reproduction instruction is input and installed in the computer device, a sound to be reproduced is stored in the computer device. A sound analysis unit that performs sound analysis on a signal and separates the sound signal into a sound signal in a sound section and a sound signal in a non-speech section.
A speech speed conversion unit that converts the speech speed so that the beginning portion is slower than a predetermined playback speed and gradually returns to the predetermined playback speed toward the end, and the speech speed conversion is performed by the speech speed conversion unit. Non-speech section adjustment for outputting, without adaptively deleting or compressing a speech signal in a non-speech section separated by the sound analysis unit based on time information of the speech signal, without shortening the speech signal to a length shorter than a predetermined length. A synthesizing unit for synthesizing the voice signal subjected to the voice speed conversion by the voice speed converting unit and the voice signal of the non-voice section obtained by the non-voice section adjusting unit to generate a converted voice signal; Is stored.

According to a fourteenth aspect of the present invention, in a recording medium storing a program for operating a computer device, when a voice reproduction instruction is input and installed in the computer device, a sound to be reproduced is stored in the computer device. A sound analysis unit that performs sound analysis on a signal and separates the signal into a sound signal in a sound section and a sound signal in a non-speech section, and calculates a fundamental frequency included in the sound signal in the sound section separated by the sound analysis unit. According to the frequency calculation unit and the change rate of the fundamental frequency obtained by the fundamental frequency calculation unit,
A speech speed conversion unit that adaptively expands the speech signal of the speech section separated by the sound analysis unit to convert a speech speed; a speech signal that has undergone speech speed conversion by the speech speed conversion unit; An audio reproduction program for synthesizing the audio signal of the non-speech section separated by the analysis unit and generating a synthesis unit for generating a converted audio signal is stored.

According to a fifteenth aspect of the present invention, in the recording medium according to the fourteenth aspect, the audio reproduction program is installed in the computer device, and when a voice reproduction instruction is input, the audio analysis program is stored in the computer device. When expanding the audio signal of the audio section separated by the section, the change rate of the fundamental frequency obtained by the fundamental frequency calculation section is compared with a preset change rate threshold, and the fundamental frequency calculation section A speech speed conversion unit for generating an expansion rate for an audio signal in a section in which the obtained change rate of the fundamental frequency exceeds a preset change rate threshold value greater than the expansion rate for the preceding and following audio signals. Features.

According to a sixteenth aspect of the present invention, in the recording medium according to the fourteenth aspect, the sound reproduction program is installed in the computer device, and when a sound reproduction instruction is input, the sound analysis program is stored in the computer device. When expanding the audio signal of the audio section separated by the section, the change rate of the fundamental frequency obtained by the fundamental frequency calculation section is compared with a preset change rate threshold, and the fundamental frequency calculation section When the obtained rate of change of the fundamental frequency exceeds a preset rate of change threshold, a certain period of time from the appearance time of the voice section, or until a certain number of voiced sound sections appear after the voice section. During this time, a speech speed conversion unit for expanding the audio signal is generated at the same expansion rate.

According to the configuration of each of the above-mentioned claims, the start portion of each sound is reproduced at a speed lower than the designated speed, and the loss of information is kept as small as possible.
While making it easier to hear the sound played at the specified speed,
High-speed playback is performed in a time frame substantially the same as the time frame required to reproduce the entire recorded audio at the specified speed, and thereby the content of the recorded audio can be searched at high speed by a program editor or the like.

[0029]

DESCRIPTION OF THE PREFERRED EMBODIMENTS <First Embodiment> FIG. 1 shows an audio reproducing apparatus and a recording medium according to the present invention.
FIG. 3 is a block diagram showing an embodiment corresponding to FIG.

As shown in FIG. 1, the audio reproducing apparatus 1a
Includes an acoustic analysis unit 2a, a speech speed conversion unit 3a, a non-speech section length control unit 4a, and a synthesis unit 5a, and separates supplied speech data into a speech section and a non-speech section. For a voice section that is dependent on a non-voice section that is longer than a certain length of time, the beginning of that section is slower than the predetermined playback speed,
The speech speed is converted so as to gradually return to the predetermined playback speed toward the end, so that important parts can be easily heard, and the designated speed (n times speed) without greatly delaying from the playback time frame.
To play the sound. The audio data supplied may be audio data recorded along with video data on a CD-RW or DVD via a communication line or the like, or audio data supplied via a communication line (main control room,
Audio data paired with video data output from the sub-control room, etc.), video signals recorded analogly on a video tape, etc., audio data obtained by digitizing audio signals, or video data on a video tape, etc. And audio data recorded digitally. The same applies to the following second to fourth embodiments.

The acoustic analysis unit 2a uses a power threshold value set in advance to perform n-times speed (n; a rational number of n> 0,
Voice data designated to be reproduced in the same manner) is divided into a voice section and a non-voice section, and the voice data included in the voice section and time information for this voice data are transmitted to the speech speed conversion section 3a. At the same time, the voice data of the non-voice section is supplied to the non-voice section length control unit 4a.

The speech speed conversion section 3a converts the speech waveform of the speech data (speech data included in the speech section) output from the sound analysis section 2a into a predetermined rule, for example, a start portion of a speech uttered in a breath. The speech speed of the voice section is always expanded by the rule that the speed is converted relatively lower than the desired playback speed, for example, n times speed, and the remaining voice is gradually converted toward the end to return to the predetermined playback speed. Is changed successively,
The obtained audio data is supplied to the synthesizing unit 5a, and the time information output from the acoustic analysis unit 2a (time information added to the audio data of the voice section)
Is compared with the time information of the voice data after the speech speed conversion to generate delay time information indicating the delay time of the voice waveform after the voice speed conversion with respect to the voice waveform before the voice speed conversion. It is supplied to the length controller 4a.

The non-speech section length control unit 4 a
a based on the delay time information output from the
The voice data output from a is adaptively deleted or compressed, and the length of the voice data included in the non-voice section is subjected to voice data expansion processing of the speech speed conversion unit 3a. In addition to the length necessary to eliminate the delay time caused by the above, the voice data of the non-voice section obtained by the deletion processing and the compression processing is supplied to the synthesis unit 5a.

The synthesizing unit 5a synthesizes the speech speed converted speech data output from the speech speed conversion unit 3a with the deleted and compressed speech data output from the non-speech section length control unit 4a. Then, the audio data obtained by the synthesis processing is supplied to a sound board (not shown), and a speaker outputs a sound with a small delay time and easy to hear.

As described above, in the first embodiment, when an instruction to reproduce audio data to be reproduced at a specified speed (n times speed) is input, the supplied audio data is divided into an audio section and a non-audio section. After that, for a voice section that is dependent on a non-voice section that is longer than a certain time length, the beginning of the voice section is slower than the predetermined playback speed, and gradually returns to the predetermined playback speed toward the end. The sound speed is converted to the sound speed, so that the important parts are easy to hear, and the sound is played back at n times speed without greatly delaying from the playback time frame. The target voice is divided into units that can be uttered in a single breath, and the start of each voice is divided at a speed slower than n times speed.
Playback, thereby reducing the loss of information as much as possible, making it easier to hear the sound reproduced at n times speed, and in the same time frame as that required to reproduce the entire recorded sound at n times speed. The audio can be reproduced at high speed, and the contents of the recorded audio can be searched at high speed by a program editor or the like (effect of claim 1).

In the first embodiment, when the recorded voice is played back at n times speed, the speech speed converter 3a controls the non-voice section length controller 4a by the length of the waveform length of the voice portion. , After shortening the length of non-speech segments,
The synthesis unit 5a adds the voice data in the voice section and the voice data in the non-voice section to generate converted voice data, so that the information is reproduced at n times speed while minimizing the loss of information. It is possible to make the length of the entire voice section and non-voice section before the speech rate conversion and the length of the entire voice section and non-voice section after the voice rate conversion substantially equal to each other while making the voice easy to hear ( Effect of Claim 2).

[Second Embodiment] FIG. 2 is a block diagram showing an embodiment of a sound reproducing apparatus and a recording medium according to the present invention, corresponding to the third and fourth aspects.

The audio reproducing apparatus 1b shown in FIG.
Non-voice determination unit 6, sound analysis unit 2b, speech speed conversion unit 3b
And a non-voice section length control unit 4b and a synthesis unit 5b.
Using a threshold value corresponding to the power value of the supplied voice data, the voice data is separated into a voice section and a non-voice section. The speech speed is converted so that the beginning portion is slower than the predetermined playback speed and gradually returns to the predetermined playback speed toward the end, so that the important portion is easily heard while the delay is greatly delayed from the playback time frame. And reproduce the sound at the specified speed (n times speed).

At this time, the voice / non-voice determination unit 6 detects the power value of the voice data designated to be reproduced at n times speed, and converts the voice data into the voice data of the voice section and the voice data of the non-voice section. In addition to generating a power threshold value necessary to separate the voice signal into a voice signal, the speech speed conversion process determines how much the voice is delayed based on the delay time information output from the non-voice section length control unit 4b. . When it is determined that the delay of the converted voice is conspicuous compared to the original voice, in order to increase the number of objects to be deleted, the ratio determined as a voice segment is reduced, while the ratio determined as a non-voice segment is increased. , Adjust the power threshold adaptively. The power threshold value thus obtained and the audio data designated to be reproduced at n times speed are supplied to the acoustic analysis unit 2b.

The sound analysis unit 2b includes a voice / non-voice determination unit 6.
The voice data output from the voice / non-voice determination unit 6 is separated into a voice section and a non-voice section using the power threshold value output from the Speech rate converter 3 converts time information for data
b, and the voice data of the non-voice section is supplied to the non-voice section length control unit 4b.

The speech speed conversion unit 3b converts the speech waveform of the speech data (speech data included in the speech section) output from the sound analysis unit 2b into a predetermined rule, for example, a start portion of a speech uttered in a breath. In accordance with the rule that the sound is converted relatively slower than the desired reproduction speed, for example, n times speed, and the remaining sound is gradually converted toward the end so as to return to the predetermined reproduction speed, the sound is expanded in the speech section. The speed is sequentially changed, and the obtained voice data is supplied to the synthesizing unit 5b, and the time information output from the acoustic analysis unit 2b (time information added to the voice data of the voice section)
Is compared with the time information of the voice data after the speech speed conversion to generate delay time information indicating the delay time of the voice waveform after the voice speed conversion with respect to the voice waveform before the voice speed conversion. It is supplied to the length control unit 4b.

The non-speech section length control unit 4 b
b) the delay time information output from the voice / non-voice determination unit 6
The voice data (voice data included in the non-voice section) output from the voice analysis unit 2b is adaptively deleted or compressed based on the delay time information to transfer the voice data included in the non-voice section. The length of the data is set to a length necessary to eliminate the delay time caused by the voice data decompression processing of the voice speed conversion unit 3b, and the voice data of the non-voice section obtained by the deletion processing and compression processing is deleted. It is supplied to the synthesizing unit 5b.

The synthesizing unit 5b synthesizes the speech speed converted speech data output from the speech speed conversion unit 3b with the deleted and compressed speech data output from the non-speech section length control unit 4b. Then, the audio data obtained by the synthesis processing is supplied to a sound board (not shown), and a speaker outputs a sound with a small delay time and easy to hear.

As described above, in the second embodiment, n
When an instruction to reproduce audio data to be played at double speed is input, the audio data is separated into an audio section and a non-audio section using a threshold value corresponding to the power value of the supplied audio data. For a voice section that is dependent on a non-voice section that is longer than a certain time length, the voice speed is converted so that the beginning is slower than a predetermined playback speed and gradually returns to the predetermined playback speed toward the end. Since the audio is played back at n times speed without significantly delaying the playback time frame while making the important parts easy to hear, the power of the sound to be played at high speed when playing the recorded sound at n times speed Using the optimal power threshold according to the value,
Even during a voice section, it is possible to efficiently delete portions considered to be of low importance in the same way as a non-voice section during listening, thereby minimizing information loss and reproducing at n times speed. At the same time frame as necessary to play the entire recorded sound at n times speed while making the recorded sound easy to hear, the sound is played at high speed and the contents of the recorded sound are edited by the program editor etc. Can be searched at high speed (the effect of claim 3).

In the second embodiment, when the recorded voice is played back at n times speed, the high-speed playback target is determined according to the delay time of the voice waveform after the voice speed conversion with respect to the voice waveform before the voice speed conversion. The power threshold required to separate the audio data into the audio data of the voice section and the voice data of the non-voice section is changed, so that not only the voice data included in the non-voice section but also the voice data included in the voice section are changed. Of the audio data, the audio part considered to be less important for listening is also deleted, so even if the non-audio section is shorter than the length of the audio section, the entire recorded audio can be deleted. The audio can be reproduced at a high speed in a time frame substantially equal to the time frame required for the reproduction at the n-times speed while maintaining the easiness of hearing (the effect of claim 4).

<< Third Embodiment >> FIG. 3 is a block diagram showing an embodiment corresponding to claim 5 of the audio reproducing apparatus and the recording medium according to the present invention.

The audio reproducing apparatus 1c shown in this figure includes an acoustic analysis unit 2c, a speech speed conversion unit 3c, a non-speech section length determination / control unit 7, and a synthesis unit 5c. The voice section and the non-voice section are separated, and the voice section which is dependent on the non-voice section of a certain time length or more is
The speech speed is converted so that the beginning portion is slower than the predetermined playback speed and gradually returns to the predetermined playback speed toward the end, and the length of the non-speech section is kept at a certain length or more, and important portions are Is reproduced at a specified speed (n times speed) without greatly delaying from the reproduction time frame while making it easy to hear.

At this time, the acoustic analysis unit 2c separates the audio data designated to be reproduced at n times speed into a voice section and a non-voice section by using a preset power threshold, and The voice data included in the section and the time information for the voice data are supplied to the speech speed conversion unit 3c, and the voice data of the non-voice section is supplied to the non-voice section length determination unit 7.

The speech speed conversion section 3c converts the speech waveform of the speech data (speech data included in the speech section) output from the sound analysis section 2c into a predetermined rule, for example, a start portion of a speech uttered in a breath. In accordance with the rule that the sound is converted relatively slower than the desired reproduction speed, for example, n times speed, and the remaining sound is gradually converted toward the end so as to return to the predetermined reproduction speed, the sound is expanded in the speech section. The speed is sequentially changed, and the obtained voice data is supplied to the synthesizing unit 5c, and the time information output from the acoustic analysis unit 2c (time information added to the voice data of the voice section)
Is compared with the time information of the voice data after the speech speed conversion to generate delay time information indicating the delay time of the voice waveform after the voice speed conversion with respect to the voice waveform before the voice speed conversion. It is supplied to the length determination / control unit 7.

When the voice data of the non-voice section is output from the voice analysis section 2c, the non-voice section length determination / control section 7 controls the voice analysis section based on the delay time information output from the speech speed conversion section 3c. The voice data (voice data included in the non-voice section) output from 2c is adaptively deleted or compressed, and the length of the voice data included in the non-voice section is processed by the voice data decompression processing of the speech speed conversion unit 3c. In order to make the length necessary to eliminate the delay time caused by the above, a very short set section length (a sound data included in a sound section
In the case of reproducing at 0 times speed, for example, the minimum length of about 100 ms is left, and the audio data of the non-voice section obtained by the deletion processing and compression processing is supplied to the synthesizing unit 5c.

The synthesizing unit 5c includes the speech speed converted speech data output from the speech speed conversion unit 3c and the speech data of the non-speech section output from the non-speech section length determination unit 7 or the non-speech section length control unit. 4c is combined with the deleted and compressed audio data, and the audio data obtained by the synthesis is supplied to a sound board (not shown). In addition, it outputs a sound that is easy to hear.

As described above, in the third embodiment, n
When an instruction to reproduce audio data to be played at double speed is input, audio data supplied via a CD-RW, a DVD, a communication line, or the like is separated into an audio section and a non-audio section, and then a predetermined period of time. For the voice section that is dependent on the non-voice section longer than the length, the voice speed is converted so that the beginning of the voice section is slower than the predetermined playback speed and gradually returns to the predetermined playback speed toward the end. The length of the section is kept at a certain length or more, and the important parts are made easier to hear, and the sound is reproduced at n times speed without greatly delaying from the reproduction time frame. When the recorded voice is played back at n times speed while being deleted, the voice to be played at high speed is divided into units that can be uttered in a single breath while leaving the utterance boundary at the start of the voice, and the Play the entire recorded sound at n times speed while playing back the start portion at a speed slower than n times speed, thereby minimizing the loss of information and making it easier to hear the sound played at n times speed. In a time frame almost the same as the time frame required for the reproduction, the audio can be reproduced at a high speed, and the contents of the recorded audio can be searched by the program editor or the like at a high speed (effect of claim 5).

<< Fourth Embodiment >> FIG. 4 is a block diagram showing an embodiment of a sound reproducing apparatus and a recording medium according to the present invention corresponding to claims 6, 7 and 8.

The audio reproducing apparatus 1d shown in this figure includes an acoustic analyzing unit 2d, a fundamental frequency calculating unit 8, a speech speed converting unit 3d
And a non-voice section length control unit 4d and a synthesis unit 5d.
After the supplied voice data is separated into a voice section and a non-voice section, the waveform is expanded adaptively according to the change in the fundamental frequency of the voice section that is dependent on the non-voice section for a certain length of time or more. To make the important parts easier to hear,
The sound is reproduced at the specified speed (n times speed) without greatly delaying from the reproduction time frame.

At this time, the sound analysis unit 2d uses the preset power threshold value while supplying the time information of the audio data designated to be reproduced at n times speed to the speech speed conversion unit 3d. , The audio data designated to be reproduced at n times speed is separated into a voice section and a non-voice section, and the voice data included in the voice section is supplied to the fundamental frequency calculating section 8 and the voice of the non-voice section is The data is supplied to the non-voice section length control unit 4d.

The fundamental frequency calculation unit 8 sequentially calculates the fundamental frequency of the speech waveform indicated by the speech data of the speech section output from the acoustic analysis unit 2d, and generates fundamental frequency information based on the calculation result. The basic frequency information and the audio data of the audio section output from the audio analysis unit 2d are supplied to the speech speed conversion unit 3d.

The speech speed converter 3d compares the temporal change rate of the fundamental frequency indicated by the fundamental frequency information output from the fundamental frequency calculator 8 with a preset change rate threshold. When the temporal change rate of the fundamental frequency is smaller than the threshold value of the change rate, the voice data (voice data included in the voice section) output from the acoustic analysis unit 2d according to the temporal change of the fundamental frequency. The speech waveform of the speech section is adaptively expanded, the speech speed of the speech section is sequentially changed, and when the temporal change rate of the fundamental frequency is larger than the change rate threshold, a predetermined rule, for example, the time of the fundamental frequency The audio waveform of the audio data (the audio data included in the audio section) output from the acoustic analysis unit 2d in the section where the dynamic change rate is larger than the change rate threshold is set at an expansion rate larger than the expansion rate of the preceding and succeeding sections. , When extended Rule or a certain period of time from the time when the temporal change rate of the fundamental frequency becomes larger than the change rate threshold (or until a certain number of voiced sections appearing after the voice section containing the time reach a certain number) In the meantime, the same expansion rate is used to expand the voice waveform by the rule of expansion, to stabilize a specific part depending on the change of the fundamental frequency, and to emphasize the part where the tone of the voice has changed. The voice data is supplied to the synthesizing unit 5d, and in parallel with these operations, the time information output from the acoustic analysis unit 2d (time information added to the voice data of the voice section) and By comparing the time information of the voice data with the time information of the voice data, delay time information indicating the delay time of the voice waveform after the voice speed conversion with respect to the voice waveform before the voice speed conversion is generated, and this is transmitted to the non-voice section length control unit 4d. To feed.

The non-speech section length control unit 4 d
d based on the delay time information output from the
d) Adaptively deletes or compresses the voice data (voice data included in the non-voice section) outputted from d, and determines the length of the voice data included in the non-voice section by the voice data expansion processing of the speech speed conversion unit 3d. In addition to the length necessary to eliminate the delay time caused by the above, the voice data of the non-voice section obtained by the deletion processing and the compression processing is supplied to the synthesis unit 5d.

The synthesizing unit 5d synthesizes the speech speed converted speech data output from the speech speed conversion unit 3d and the deleted and compressed speech data output from the non-speech section length control unit 4d. Then, the audio data obtained by the synthesis processing is supplied to a sound board (not shown), and a speaker outputs a sound with a small delay time and easy to hear.

As described above, in the fourth embodiment, n
When an instruction to reproduce audio data to be played at double speed is input, the supplied audio data is separated into an audio section and a non-audio section, and then the audio section that is interrupted by the non-audio section that is longer than a certain time length On the other hand, since the waveform is expanded adaptively according to the fluctuation of the fundamental frequency, the sound is reproduced at n times speed without greatly delaying from the reproduction time frame while the important part is easily heard. When playing back recorded voice at n times speed, the voice to be played at high speed is divided into units that can be uttered in one breath, and the portion where the fundamental frequency of each voice fluctuates is played back at a speed slower than n times speed. It is possible to focus on the part where the pitch of the voice has changed, and to reduce the loss of information as much as possible. The audio can be played back at high speed in a time frame substantially equal to the time frame required for playing back the entire audio at n times speed, and the contents of the recorded audio can be searched at high speed by a program editor or the like. 6 effect).

In the fourth embodiment, as a rule to be applied when the temporal change rate of the fundamental frequency is larger than the change rate threshold, the temporal change rate of the fundamental frequency is It is possible to select a rule to extend the audio waveform of the audio data (the audio data included in the audio section) output from the acoustic analysis unit 2d at an expansion rate larger than that of the preceding and succeeding sections only in the larger section. When the recorded sound is played back at n times speed, the part where the fundamental frequency of each sound fluctuates greatly can be played back at a speed slower than the surrounding speed, and this allows the information to be reproduced. While the loss of voices is kept as small as possible, the portion where the voice pitch has changed is emphasized and the voice reproduced at the n-times speed can be easily heard. The effect of section 7).

In the fourth embodiment, as a rule to be applied when the temporal change rate of the fundamental frequency is larger than the change rate threshold, the temporal change rate of the fundamental frequency is Select a rule to extend at the same decompression rate for a certain period of time (or until a certain number of voiced sections appearing after the voice section containing the time reaches a certain number) from the time when the time becomes larger. When the recorded voice is played back at n times speed, a portion where the fundamental frequency of each voice fluctuates greatly is replaced by a voice segment following the voiced voice segment for a certain period of time or a certain number of voiced voice segments. Until the appearance of the sound, it can be played back at the same slow speed, thereby keeping the loss of information as small as possible, It was allowed to focus on extending between, can be easier to hear the sound reproduced by the n-times speed (effect of claim 8).

<< Other Embodiments >> FIG. 5 is a block diagram showing an example in which a recording medium on which an audio reproduction program according to the present invention is recorded is installed in a computer to constitute the audio reproduction apparatus shown in FIGS. It is.

That is, the computer in which the sound reproducing program 13 stored in the recording medium 11 is installed is used by the sound analyzers 2a to 2d and the speech speed converter 3
a to 3d, non-voice section length control units 4a to 4d, synthesis unit 5a
To 5d, a voice / non-voice determining unit 6, a non-voice section length determining unit 7, and a voice reproducing device 1a in which a fundamental frequency calculating unit 8 is generated.
1d, the keyboard and the mouse of the computer device are operated to input an instruction to reproduce audio data to be reproduced at n-times speed.
5, audio data supplied via the CD-RW 16, DVD 17, etc., is separated into a voice section and a non-voice section. Part is slower than the specified playback speed,
In addition, the speech speed is converted so as to gradually return to a predetermined playback speed toward the end, and the audio is played back at n times speed without greatly delaying from the playback time frame while making important parts easy to hear.

In this case, the recording medium 11 is a CD-RO
Recording medium body 12 composed of M, DVD, etc.
And a sound reproducing program 13 recorded on the recording medium main body 12, and a setup program 14 for setting the sound reproducing program 13 in the computer main body to construct the sound reproducing devices 1a to 1d.
When an installation instruction is input, the recording medium body 12
The audio reproduction program 13 stored in the recording medium main body 12 is transferred to the CPU of the computer main body by the setup program 14 stored in the hard disk mechanism, and is installed in the hard disk mechanism. Thus, each of the audio reproducing apparatuses 1a to 1a shown in FIGS.
1d can be configured.

In the above example, the audio reproducing devices 1a to 1a
As d, the sound reproduction program 13 stored in the recording medium 11 is installed in the computer device, and the sound analysis units 2a to 2d, the speech speed conversion units 3a to 3d, the non-speech section length control units 4a to 4d, and the synthesis unit 5a to 5d, a voice / non-voice determination unit 6, a non-voice section length determination unit 7, and a fundamental frequency calculation unit 8 are generated. However, LSI devices, IC devices, transistor devices, resistors, capacitors, coils, etc. Using discrete components, these acoustic analysis units 2
a to 2d, speech speed converters 3a to 3d, non-speech section length controllers 4a to 4d, synthesis sections 5a to 5d, speech / non-speech decision section 6, non-speech section length decision section 7, and fundamental frequency calculation section 8. Alternatively, a sound reproduction device may be created.

Then, by using the audio reproducing device created in this way, the speech speed converting process can be performed faster and more efficiently than the audio reproducing devices 1a to 1d using the computer device. The efficiency of the program editing work can be further improved.

[0068]

As described above, according to the present invention, according to the first and ninth aspects, when a recorded voice is reproduced at a specified speed (n times speed), the voice to be reproduced at high speed is uttered in a breath. By splitting the sound into startable units and playing back the start of each sound at a speed slower than the specified speed, the loss of information is kept as small as possible, making it easier to hear the sound played at the specified speed, High-speed playback can be performed in the same time frame as that required to play the entire recorded audio at the specified speed, which allows program editors etc. to search the contents of the recorded audio at high speed Can be.

According to the second and tenth aspects, when the recorded voice is reproduced at the specified speed, the voice to be reproduced at high speed is divided into units that can be uttered in a single breath, and the start portion of each voice is set to be lower than the specified speed. By playing back at a slow speed and efficiently deleting non-sound parts, information loss is kept as small as possible, making it easier to hear the sound played at the specified speed, High-speed reproduction can be performed in a time frame substantially the same as the time frame required for reproduction at the specified speed, and thereby the contents of the recorded audio can be searched at high speed by a program editor or the like.

According to the third and eleventh aspects, when the recorded audio is reproduced at a specified speed, the optimum power threshold value according to the power value of the audio to be reproduced at high speed is used, and Is divided into units that can be uttered in a single breath, and the start of each voice is played back at a speed slower than the specified speed, and voice parts and non-voice parts that are considered to be less important for listening Efficiently removes the information loss as much as possible, making it easier to hear the sound played at the specified speed, while maintaining the time frame required to play the entire recorded sound at the specified speed. In the same time frame,
High-speed reproduction can be performed, and thereby, the contents of the recorded audio can be searched at high speed by a program editor or the like.

According to the fourth and twelfth aspects, when the recorded sound is reproduced at the specified speed, the optimum power threshold value corresponding to the power value of the sound to be reproduced at high speed is used to determine whether the sound is to be reproduced at high speed. Is divided into units that can be uttered in a single breath, and the start of each voice is played back at a speed slower than the specified speed, and voice parts and non-voice parts that are considered to be less important for listening The time frame required to play the entire recorded sound at the specified speed while reducing the loss of information as much as possible, making it easier to hear the sound played at the specified speed High-speed playback can be performed in almost the same time frame, and thereby, the contents of the recorded audio can be searched at high speed by a program editor or the like.

According to the fifth and thirteenth aspects, when the recorded voice is reproduced at a specified speed while deleting the non-voice section of the recorded voice, the high speed is maintained while leaving a vocal boundary at the start of the voice. The audio to be played is divided into units that can be uttered in a single breath, and the start of each audio is played back at a speed slower than the specified speed, thereby minimizing information loss and playing at the specified speed. The recorded audio can be played back at high speed in a time frame almost the same as that required to play the entire recorded audio at the specified speed while making it easier to hear the recorded audio. Editors can search at high speed.

According to claim 6 and claim 14, when the recorded voice is reproduced at a specified speed, the voice to be reproduced at high speed is divided into units that can be uttered in a single breath, and the portion where the fundamental frequency of each voice fluctuates is reduced. By playing back at a speed slower than the specified speed, the part where the voice pitch has changed is emphasized, thereby reducing the loss of information as much as possible and listening to the sound played at the specified speed To make it easy to play back at high speed in the same time frame as that required to play the entire recorded audio at the specified speed, and to have the program editor search the contents of the recorded audio at high speed Can be.

According to the seventh and fifteenth aspects, when a recorded voice is reproduced at a designated speed, the voice to be reproduced at a high speed is divided into units that can be uttered in a single breath, and a portion where the fundamental frequency of each voice greatly fluctuates. Was played back at a slower speed than the surrounding speed, so that the part where the voice pitch changed was emphasized, thereby minimizing information loss and playing back at the specified speed. At the same time frame as required to play the entire recorded sound at the specified speed while making the sound easy to hear, it is played back at high speed, and the contents of the recorded sound are sent to the program editor etc. at high speed. Can be searched.

According to the eighth and sixteenth aspects, when the recorded voice is reproduced at a specified speed, the voice to be reproduced at high speed is divided into units that can be uttered in a single breath, and a portion where the fundamental frequency of each voice greatly fluctuates. The voice pitch was changed by playing the voice section following the voiced section for a certain period of time or at the same slow speed until a certain number of voiced sections appeared. The time frame required to play the entire recorded sound at the specified speed while focusing on the part, thereby minimizing the loss of information and making the sound played at the specified speed easier to hear In almost the same time frame, high-speed reproduction can be performed, and the contents of the recorded audio can be searched at high speed by a program editor or the like.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment corresponding to claims 1, 2, 9, and 10 of an audio reproducing apparatus and a recording medium according to the present invention.

FIG. 2 is a block diagram showing an embodiment corresponding to claims 3, 4, 11, and 12 of the audio reproducing apparatus and the recording medium according to the present invention.

FIG. 3 is a block diagram showing an embodiment corresponding to claims 5 and 13 of the audio reproducing apparatus and the recording medium according to the present invention.

FIG. 4 is a block diagram showing an embodiment corresponding to claims 6, 7, 8, 14, 15, and 16 of the audio reproducing apparatus and the recording medium according to the present invention.

FIG. 5 is a block diagram showing an example in which a recording medium on which an audio reproduction program according to the present invention is recorded is installed in a computer device to constitute the audio reproduction device shown in FIGS.

[Explanation of symbols]

 1a to 1d: voice reproducing device 2a to 2d: sound analysis unit 3a to 3d: speech speed conversion unit 4a, 4b, 4d: non-voice section length control unit 5a to 5d: synthesis unit 6: voice / non-voice determination unit 7: Non-voice section length determination / control section (non-voice section adjustment section) 8: fundamental frequency calculation section 11: recording medium 12: recording medium body 13: voice reproduction program 14: setup program 15: communication line 16: CD-RW 17: DVD

Continued on the front page (72) Inventor Toru Toki 1-10-11 Kinuta, Setagaya-ku, Tokyo F-term in the Japan Broadcasting Corporation Research Institute of Broadcasting Technology (Reference) 5D045 BA02 9A001 BB02 BB03 BB04 DD11 EE02 EE04 FF03 HH16 HH18 JJ72 KK43 KK60

Claims (16)

[Claims] 1. A sound analysis unit that performs sound analysis on a sound signal to be reproduced and separates the sound signal into a sound signal in a sound section and a sound signal in a non-speech section, and a sound in the sound section separated by the sound analysis unit. A speech speed conversion unit for converting the speech speed of the signal so that the beginning of the signal is slower than a predetermined reproduction speed and gradually returning to the end to the predetermined reproduction speed toward the end; An audio reproducing apparatus, comprising: a synthesizing unit that synthesizes a converted audio signal and an audio signal in a non-speech section separated by the acoustic analysis unit to generate a converted audio signal. 2. The sound reproducing apparatus according to claim 1, wherein the sound of the non-speech section separated by the sound analysis unit based on time information of the sound signal subjected to the speech speed conversion by the speech speed conversion unit. A sound reproducing apparatus, comprising: a non-speech section length control unit that adaptively deletes or compresses a signal and synthesizes the signal with the sound signal after the speech speed conversion in the synthesis unit. 3. A sound for determining a power value of a sound signal to be reproduced and generating a power threshold necessary for separating the sound signal into a sound signal in a sound section and a sound signal in a non-speech section. A non-speech determining unit, and a sound that separates a sound signal to be reproduced into a voice signal in a voice section and a voice signal in a non-voice section using the power threshold value obtained by the voice / non-voice determining unit. An analysis unit, for the audio signal of the audio section separated by the acoustic analysis unit, the speech speed is set so that the beginning portion thereof is slower than a predetermined reproduction speed, and gradually returns to the predetermined reproduction speed toward the end. A speech speed conversion unit to be converted; and a speech signal subjected to speech speed conversion by the speech speed conversion unit and a speech signal in a non-speech section separated by the acoustic analysis unit to generate a converted speech signal. And a combining unit. Audio playback device. 4. The audio reproduction apparatus according to claim 3, wherein the audio / non-speech determination unit separates the audio signal to be reproduced into an audio signal in an audio section and an audio signal in a non-audio section. When generating the required power threshold,
An audio reproducing apparatus characterized in that the power threshold value is adaptively changed in proportion to an amount of delay time accumulation from an original audio due to extension of an audio signal in an audio section. 5. A sound analysis unit that performs sound analysis of a sound signal to be reproduced and separates the sound signal into a sound signal in a sound section and a sound signal in a non-speech section, and a sound in the sound section separated by the sound analysis unit. A speech speed conversion unit for converting the speech speed of the signal so that the beginning of the signal is slower than a predetermined reproduction speed and gradually returning to the end to the predetermined reproduction speed toward the end; Based on the time information of the converted audio signal, when adaptively deleting or compressing the audio signal of the non-speech section separated by the acoustic analysis unit,
A non-speech section adjusting unit that outputs a voice signal without being shorter than a predetermined length; a speech signal whose speech rate has been converted by the speech rate conversion unit; and a speech signal of a non-speech section obtained by the non-speech section adjustment unit. And a synthesizing unit for synthesizing and generating a converted audio signal. 6. A sound analysis unit for sound-analyzing a sound signal to be reproduced and separating the sound signal into a sound signal in a sound section and a sound signal in a non-speech section, and a sound in the sound section separated by the sound analysis unit. A fundamental frequency calculator for calculating a fundamental frequency included in the signal, and adaptively converting a speech signal of a speech section separated by the acoustic analyzer in accordance with a change rate of the fundamental frequency obtained by the fundamental frequency calculator. A speech speed conversion unit for extending the speech speed and converting the speech speed; and synthesizing a speech signal subjected to speech speed conversion by the speech speed conversion unit and a speech signal of a non-speech section separated by the acoustic analysis unit. And a synthesizer for generating a converted audio signal. 7. The sound reproduction device according to claim 6, wherein the speech speed conversion unit expands a sound signal in a sound section separated by the sound analysis unit, and obtains a basic signal obtained by a fundamental frequency calculation unit. A section in which the change rate of the frequency is compared with a preset change rate threshold value, and the change rate of the fundamental frequency obtained by the fundamental frequency calculation section exceeds the preset change rate threshold value. A decompression ratio for the audio signal of the first and second audio signals is made larger than a decompression ratio for the preceding and succeeding audio signals. 8. The sound reproduction device according to claim 6, wherein the speech speed conversion unit expands a speech signal in a speech section separated by the sound analysis unit, and obtains a basic signal obtained by a fundamental frequency calculation unit. The change rate of the frequency is compared with a preset change rate threshold, and when the change rate of the fundamental frequency obtained by the fundamental frequency calculation unit exceeds the preset change rate threshold, An audio reproducing apparatus characterized in that an audio signal is expanded at the same expansion rate for a fixed time from the appearance time of the voice section or until a certain number of voiced sound sections appear after the voice section. 9. A recording medium in which a program for operating a computer device is stored, wherein when installed in the computer device and an audio reproduction instruction is input, an audio signal to be reproduced is sounded in the computer device. An audio analysis unit that analyzes and separates the audio signal into an audio signal in a voice section and an audio signal in a non-voice section; And a speech speed conversion unit that changes the speech speed so as to gradually return to a predetermined playback speed toward the end, and a speech signal whose speech speed has been converted by the speech speed conversion unit and the acoustic analysis unit. A recording medium, comprising: a synthesizing unit for synthesizing a separated non-voice section voice signal to generate a converted voice signal; and a voice reproduction program for generating a voice signal. 10. The recording medium according to claim 9, wherein the audio reproduction program is installed in the computer device, and when a voice reproduction instruction is input, the audio reproduction program is installed in the computer device by the speech speed conversion unit. Based on the time information of the speech signal subjected to the speech rate conversion, the speech signal of the non-speech section separated by the acoustic analysis unit is adaptively deleted or compressed, and the speech rate after the speech rate conversion is performed by the synthesis unit. A non-voice section length control unit for synthesizing with a voice signal is generated. 11. A recording medium in which a program for operating a computer device is stored, wherein the power of an audio signal to be reproduced is installed in the computer device when an audio reproduction instruction is input when the program is installed in the computer device. A voice threshold for generating a power threshold required to separate the voice signal into a voice signal in a voice section and a voice signal in a non-voice section by determining the voice signal.
A non-speech determining unit, and an audio analysis for separating a voice signal to be reproduced into a voice signal in a voice section and a voice signal in a non-voice section using a power threshold obtained by the voice / non-voice determining unit. And the voice signal of the voice section separated by the acoustic analysis unit, the speech speed is converted so that the beginning is slower than a predetermined playback speed and gradually returns to the predetermined playback speed toward the end. A speech rate conversion unit that performs speech rate conversion by the speech rate conversion unit and a speech signal in a non-speech section separated by the acoustic analysis unit to generate a converted speech signal. A recording medium characterized by storing an audio reproducing program for generating a sound reproducing program. 12. The recording medium according to claim 11, wherein the audio reproduction program is installed in the computer device, and when an audio reproduction instruction is input, an audio signal to be reproduced is stored in the computer device. Generates a power threshold required to separate the audio signal into the audio signal of the voice section and the voice signal of the non-voice section, and stores the delay time from the original voice due to the expansion of the voice signal in the voice section. A recording medium characterized by generating the voice / non-voice determination unit that adaptively changes the power threshold in proportion to an amount. 13. A recording medium in which a program for operating a computer device is stored. When a sound reproduction instruction is input and installed in the computer device, an audio signal to be reproduced is sounded in the computer device. An audio analysis unit that analyzes and separates the audio signal into an audio signal in a voice section and an audio signal in a non-voice section; And a speech speed conversion unit that changes the speech speed so as to gradually return to a predetermined playback speed toward the end, based on time information of the audio signal subjected to the speech speed conversion by the speech speed conversion unit, When adaptively deleting or compressing the audio signal of the non-speech section separated by the acoustic analysis unit,
A non-speech section adjustment unit that outputs a voice signal without being shorter than a predetermined length; a speech signal whose speech rate has been converted by the speech rate conversion unit; and a speech signal of a non-speech section obtained by the non-speech section adjustment unit. And a synthesizing unit for synthesizing and generating a converted audio signal, and a sound reproducing program for generating an audio signal. 14. A recording medium in which a program for operating a computer device is stored, wherein when the computer device is installed and an audio reproduction instruction is input, an audio signal to be reproduced is sounded in the computer device. An acoustic analysis unit that analyzes and separates the audio signal into a voice section and a non-voice section voice signal; and a basic frequency calculation unit that calculates a fundamental frequency included in the voice signal of the voice section separated by the acoustic analysis section. And a speech speed conversion unit that adaptively expands the speech signal of the speech section separated by the acoustic analysis unit and converts the speech speed according to the change rate of the fundamental frequency obtained by the fundamental frequency calculation unit. And synthesizing the voice signal subjected to the voice speed conversion by the voice speed conversion unit and the voice signal of the non-voice section separated by the acoustic analysis unit to generate a converted voice signal. Recording medium, wherein the audio playback program to generate a generating unit, is stored. 15. The recording medium according to claim 14, wherein the sound reproduction program is installed in the computer device, and is separated by the sound analysis unit in the computer device when a sound reproduction instruction is input. When expanding the voice signal of the voice section, the rate of change of the fundamental frequency obtained by the fundamental frequency calculation unit is compared with a preset change rate threshold, and obtained by the fundamental frequency calculation unit. Generating a speech speed conversion unit that makes the expansion rate for an audio signal in a section where the change rate of the fundamental frequency exceeds a preset change rate threshold larger than the expansion rate for the preceding and following audio signals; Recording medium. 16. The recording medium according to claim 14, wherein the sound reproduction program is installed in the computer device, and is separated by the sound analysis unit in the computer device when a sound reproduction instruction is input. When expanding the voice signal of the voice section, the rate of change of the fundamental frequency obtained by the fundamental frequency calculation unit is compared with a preset change rate threshold, and obtained by the fundamental frequency calculation unit. When the rate of change of the fundamental frequency exceeds a preset rate of change threshold, a certain time from the appearance time of the voice section, or a time until a certain number of voiced sound sections appear after the voice section, A recording medium characterized by generating a speech speed conversion unit for expanding an audio signal at the same expansion rate.