JP3327936B2 - Speech rate control type hearing aid - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to the processing of speech of a speaker in audio equipment and medical equipment such as a television, a radio tape recorder and a hearing aid, for example, to increase the sound speed of the listener in almost real time. Fit,
The present invention relates to a speech rate control type hearing aid for performing hearing aid.

[Summary of the Invention] [0002] The present invention relates to a speech rate control type hearing aid for compensating for a listening ability such as a critical voice discrimination speed (a maximum speech speed at which a voice can be accurately discriminated) which decreases due to aging or some obstacle. The present invention relates to a method and an apparatus for converting a speech speed to a high quality by a listener's own operation while maintaining a speaker's personality and phonological characteristics, thereby obtaining an optimal speech intelligibility for the listener.

[0003]

2. Description of the Related Art In general, when a listener loses his / her hearing ability such as a critical speed for speech recognition (maximum speech speed at which speech can be accurately identified) due to aging or some obstacle, for example, a normal speed speech or a quick speech is used. The discrimination degree of the spoken voice is greatly reduced.

[0004]

However, conventionally, only hearing aids have been known as hearing aids for persons with such hearing impairment, and hearing aids are provided by simply improving the frequency characteristics and controlling the gain. Since only the transfer characteristics of the outer and middle ears of the system were compensated, it was not possible to compensate for the decrease in the ability to discriminate speech mainly involving the deterioration of the auditory center.

[0005] It is an object of the present invention to compensate for hearing impairment mainly due to the auditory center by converting the speaking rate into a high quality one in near real time so as to optimize the listening ability of the listener. It is an object of the present invention to provide a speech rate control type hearing aid that can be used.

[0006]

The speech rate control type hearing aid of the present invention is characterized in that each of the following modules (1) to (10) is constituted by a transputer module comprising an IC for parallel operation. I have. That is, (1)
An A / D conversion module for A / D converting an input audio signal to generate time-series audio data; and (2) buffering for taking in the A / D-converted time-series audio data and processing it in frame units And (3) calculate an average power, a number of zero crossings, and an autocorrelation coefficient with respect to the time-series audio data taken in through the input buffer module, and calculate a silent frame based on these thresholds. A first analysis module that executes a section division process for determining unvoiced frames and voiced frames, and (4) a second analysis module that performs decimation for speeding up the time-series voice data that has been section-divided. (5) calculating an autocorrelation coefficient of the decimated time-series audio data and obtaining a pitch frequency of the audio for each frame; An analysis module; and (6) a fourth analysis module for smoothing the trajectory of the determined pitch frequency to determine the starting point and the number of pitches and for performing pitch section length extension and repetition processing, and (7) determination. A fifth section that determines each pitch section of a silent section, an unvoiced section, and a voiced section based on the starting point of the pitch and the number of pitches
And (8) converting the speech speed so as to conform to the extension ratio of the silent section, the unvoiced section, the voiced section, and the unvoiced section and the voiced section obtained from the voice speed parameter setting means obtained by the fifth analysis module. (9) an output buffer module that performs buffering for maintaining continuity of the speech synthesis output of the synthesis module, and (10) a D / A conversion of the output of the output buffer module. And a D / A conversion module for outputting a speech signal whose speech speed is controlled by a transputer module.

[0007]

[0008]

According to the speech rate control type hearing aid of the present invention, first, the input audio signal is converted into an A / D signal by the A / D conversion module.
It is converted to generate time-series audio data. The buffer module takes in the A / D converted time-series audio data and performs buffering for processing on a frame-by-frame basis. In the first analysis module, the average power, the number of zero crossings, and the autocorrelation coefficient are calculated for the time-series audio data captured via the input buffer module, and these thresholds are used to calculate a silence frame, an unvoiced frame, and a voiced frame. Section division processing for determining a frame is executed. In the second analysis module, decimation for speeding up is performed on the time-series audio data divided into sections. In the third analysis module, an autocorrelation coefficient of the decimated time-series audio data is obtained,
The pitch frequency of speech is determined for each frame. In the fourth analysis module, the trajectory of the determined pitch frequency is smoothed to determine the starting point of the pitch and the number of pitches, and the pitch section length extension / repetition processing is performed. In the fifth analysis module, based on the determined start point of the pitch and the number of pitches, each pitch section of a silent section, an unvoiced section, and a voiced section is determined. In the synthesis module, the speech rate is converted to match the extension ratio of the silent section, unvoiced section, and voiced section obtained by the fifth analysis module, and the extension ratio of the silent section and the voiced section obtained from the speech rate parameter setting module. Synthesis is performed. In the output buffer module, buffering is performed to absorb the extension due to voice expansion in the voice synthesis output of the synthesis module and maintain continuity. The output of the output buffer module is D / A converted by the D / A conversion module and output as an audio signal whose speech speed is controlled.

[0009]

[0010]

[0011]

[0012]

[0013]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of the present invention will be described below in detail with reference to the drawings.

FIG. 1 is a functional block diagram of a speaking speed control type hearing aid. This apparatus comprises a section dividing section 1, a silence section extension processing section 2, a pitch period extraction section 3, a pitch section division processing section 4, a pitch section length extension / repetition processing section 5, a speech speed setting section 6, and a synthesis section 7. Have been.

The section dividing section 1 is a section for performing a section dividing process for dividing an input voice into a silent section, an unvoiced section and a voiced section.

The silent section extension processing section 2 is a section for extending the silent section divided by the section dividing section 1 at a ratio determined by the speech speed setting section 6 in advance.

The pitch period extraction processing section 3 is a section for extracting a pitch period from the voiced section divided by the section division processing section 1. The voice section is divided for each pitch section. The pitch section length extension / repetition section 5 extends each pitch section length divided by the pitch section division processing section at a ratio predetermined by the speech speed setting section 6. It is the part to be done.

The speech speed setting section 6 is a portion for setting the extension ratio of the silent section and the extension ratio of the pitch section of the voiced section in accordance with the speaking speed of the speaker and the listening ability of the listener.

[0019] The synthesizing unit 7, the extension processing of decorated with silent interval, nothing no voice interval which was not subjected to treatment, and the voiced segment having undergone extension processing of the pitch period in the same order as the input speech again This is the part that synthesizes and outputs as speech.

Next, the operation of the speaking speed control type hearing aid of the above configuration will be described.

For an input voice, first, a section division processing section 1
Is divided into a silent section i, a silent section ii, and a voiced section iii.

Then, the silent section extension processing section 2 performs extension processing on the silent section i at a predetermined ratio determined by the speech speed setting section 6, and performs control during talking.

For the voiced section iii, first, the pitch period of the voiced section iii is extracted by the pitch period extraction processing section 3, and subsequently, the pitch period is extracted by the pitch period extraction processing section 3 by the pitch section division processing section 4. The pitch section is divided according to the pitch cycle. Furthermore, pitch section length extension
The voice section iii is extended by repeating the pitch sections divided by the pitch section division processing section 4 by the repetition processing section 5 by a number predetermined by the speech speed setting section 6.

No processing is performed on the unvoiced section ii in order to maintain the speaker's personality and phonological property.

The silent section i, the unvoiced section ii, and the voiced section iii which have been processed in each section in this way are re-synthesized in the same order as the input voice in the synthesizer 7 and output as voice. In this way, the output voice has a speech speed converted according to the listening ability of the listener.

Next, a speech speed control type hearing aid according to an embodiment of the present invention will be described with reference to FIG.

Normally, the speech rate control type hearing aid shown in FIG. 1 is realized on a large computer, but in this case, a large amount of computation is required, and a long time is required for signal processing.
In addition to lack of real-time properties, it lacks portability and operability. For this reason, the duration of the input voice necessarily depends on the memory capacity, and must be terminated at a finite length.

In this embodiment, a plurality of ultra-small transputer modules (parallel operation ICs) are connected in cascade, and a signal processing algorithm is optimally distributed to each module to realize high-speed processing. Is greatly shortened to achieve real-time performance. Accordingly, this enables not only processing even when the input voice continues unlimitedly, but also small size, light weight, and excellent portability. In addition, the user himself / herself can change parameters necessary for speech speed conversion (extension ratio of the silent section i and the voiced section iii) by a simple manual operation, and can set an optimal value for the user, so that operability is improved. It can be excellent.

In the block diagram shown in FIG. 2, each block corresponds to one transputer module (TRPM), and each algorithm cascades high-speed operation processing on each module.

The speech rate control type hearing aid of this embodiment has an A /
D conversion module 11, input buffer module 12,
It comprises first to fifth analysis modules 13 to 17, a rotary encoder 18, a synthesis module 19, an output buffer module 20, and a D / A conversion module 21.

The A / D conversion module 11 quantizes the input audio signal by 16 bits and performs A / D conversion at 48 kHz sampling.
D-convert.

The input buffer module 12 sequentially takes in A / D-converted time-series audio data, and performs buffering for processing in frame units.

The first analysis module 13 which is in charge of the section division processing calculates the average power, the number of zero crossings, and the autocorrelation coefficient, and uses these thresholds to calculate the silence, unvoiced and voiced sound frames i, ii, Determine iii.

The second analysis module 14 performs decimation for speeding up.

The third analysis module 15 which is in charge of the pitch period extracting process obtains the autocorrelation coefficient of the decimated time-series audio data, and uses this to calculate the pitch frequency of the audio for each frame.

The fourth analysis module 16 in charge of the pitch section length extension and repetition processing smoothes the pitch frequency trajectory and determines the starting point of the pitch and the number of pitches.

The fifth analysis module 17 makes fine adjustments such as final error correction of the parameters obtained in each block, and determines a silent section i, an unvoiced section ii, a voiced section iii and each pitch section.

The rotary encoder 18 quantizes, as angle information, 8-bit parameters necessary for speech speed conversion (extension ratio of a silent section and a voiced section) given by manual operation.

Compositing module 19 responsible for compositing processing
Is a silent section i obtained from the fifth analysis module 17,
Speech synthesis is performed by converting the speech speed to match the extension ratio of the unvoiced section ii, the voiced section iii, and the unvoiced and voiced sections obtained from the rotary encoder 18.

Since the synthesized voice is expanded, the output buffer module 20 performs buffering to absorb the extended portion.

The D / A conversion module 21 performs D / A conversion on the synthesized voice and outputs the voice as a voice signal.

As described above, according to the speech rate control type hearing aid of this embodiment, the input voice is subjected to the section dividing process for dividing the input speech into the silent section, the unvoiced section and the voiced section, and the predetermined ratio is determined for the silent section. No processing is applied to unvoiced sections, and pitch periods are extracted for voiced sections, divided for each pitch section, and each pitch section length is extended by a predetermined ratio. Re-synthesize the silence section with extension processing, the unvoiced section without any processing, and the voiced section with extension processing of the pitch section in the same order as the input voice, according to the listening ability of the listener. It can be converted and output in near real-time to the spoken speed.

[0043]

As described above, according to the present invention, a process of converting the speech speed to a high quality while maintaining the speaker's individuality and phonological characteristics is performed, and the listener's own operation is performed by the listener's own operation. Speaking speed can be fitted to the ability, and it can be converted to the optimum talking speed for the listener almost in real time, so that the listening ability of the listener whose voice identification critical speed has decreased due to aging or some obstacle, etc., is compensated, You can listen in real time at the optimal speed.

[Brief description of the drawings]

FIG. 1 is a functional block diagram of a speech speed control type hearing aid.

FIG. 2 is a functional block diagram of a speaking speed control type hearing aid according to an embodiment of the present invention;

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 section division processing section 2 silent section extension processing section 3 pitch period extraction processing section 4 pitch section division processing section 5 pitch section length extension / repetition processing section 6 speech speed setting section 7 synthesis section 11 A / D conversion module 12 input buffer module 13 First Analysis Module 14 Second Analysis Module 15 Third Analysis Module 16 Fourth Analysis Module 17 Fifth Analysis Module 18 Rotary Encoder 19 Synthesis Module 20 Output Buffer Module 21 D / A Conversion Module

──────────────────────────────────────────────────続 き Continuation of front page (72) Inventor Eiichi Miyasaka 1-10-11 Kinuta, Setagaya-ku, Tokyo Japan Broadcasting Corporation Research Institute of Broadcasting Technology (56) References JP-A-1-93795 (JP, A) JP-A-59-82608 (JP, A) JP-A-64-44995 (JP, A) JP-A-5-73089 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G10L 21 / 04 A61F 11/00 H03M 1/00 H04R 25/00

Claims (1)

(57) [Claims] 1. A speech rate control type hearing aid wherein each of the following modules (1) to (10) is constituted by a transputer module comprising a parallel operation IC. (1) an A / D conversion module for A / D converting an input audio signal to generate time-series audio data; and (2) an A / D conversion module for taking in the A / D-converted time-series audio data and processing it in frame units. (3) An average power, a number of zero crossings, and an autocorrelation coefficient are calculated for the time-series audio data taken in through the input buffer module, and silence is performed based on these threshold values. A first analysis module that executes a section division process for determining a frame, an unvoiced frame, and a voiced frame; (4) a second analysis module that performs decimation for speeding up the time-divided time-series speech data; 3) calculating the autocorrelation coefficient of the decimated time-series audio data, and calculating the pitch frequency of the audio for each frame. An analysis module, (6) a fourth analysis module for smoothing the trajectory of the determined pitch frequency to determine the starting point and the number of pitches, and for performing pitch section length extension and repetition processing; (7) determined A fifth analysis module for determining each of the silent sections, unvoiced sections, and voiced sections based on the starting point of the pitch and the number of pitches; (8) a silent section, a unvoiced section, and a voiced section obtained by the fifth analyzing module; A synthesizing module that converts the speech speed so as to conform to the extension ratio of the silent section and the voiced section obtained from the section and the speech rate parameter setting means, and performs speech synthesis; (9) maintaining continuity in speech synthesis output of the synthesis module Output buffer module that performs buffering for performing (10) D / A conversion of the output of the output buffer module D / A conversion module for outputting a sound signal speech speed is controlled.