JPS5817499A - Variable speed type voice reproducer - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention provides an audio output close to spoken words, regardless of the playback speed, during playback by an audio playback device such as a V'l'l (video tape recorder).

Variable speed voice playback device 1 that performs voice synthesis using voice synthesis
related.

Conventionally, when playing at a speed that is different from the rated speed of the V-telephone, for example, 2x speed, o, ix speed, the frequency of the audio signal changes accordingly, so if you output the audio as is, it will not be possible. It was impossible. For this reason, during playback, the sound playback output circuit is shut off, and either the f-voice is not played back, or the sound is returned to the same wave number as the playback sound.

A method is adopted in which a part of the audio waveform is cut or the waveform is expanded.

When playing Ma'I'R at a speed different from the rated speed, the added value increases if not only one screen is played but also the sound is played. For example, when performing fast forwarding, there is a drawback in that the user must keep their eyes on one screen when audio is not being played back.

When outputting audio, methods of cutting part of the audio waveform or expanding the waveform have the disadvantage that unnatural disturbances occur in the audio waveform and the clarity of the reproduced sound is low. It was. In the above method, the playback speed is 0.5 times ~
The drawback is that it is limited to about twice as much.

The present invention has been developed in consideration of the above circumstances.

DemRCORcl autocorrelation is applied to the audio output section of the audio playback device; Partial mutoc@rr@1stIo
m) With an analytical synthesizer and during non-rated speed regeneration.

By changing the sampling frequency of the audio waveform in proportion to the playback speed, and changing the usage time of the same data in inverse proportion to the playback speed during voice synthesis, synthesis that is close to normal spoken language can be achieved regardless of the playback speed. An object of the present invention is to provide a variable speed audio reproduction device capable of outputting sound.

For example, when playing V'l'l at the rated speed.

The reproduced sound is output from the speaker as it is, but if this is not the case, the reproduced sound is not output from the speaker as is, but the audio waveform, which is analog data, is processed using a chimp clock for each frequency proportional to the playback speed.・Digital) Convert it to digital data and store it in Memo Two. Regarding this digital data,
Perform PARCOI analysis, and from the waveform domain data, P
Convert to AICO1l data.

The PAllCO1l data is composed of parameters such as pitch, amplitude, and voiced/unvoiced distinction.

Next, audio is synthesized using these PARCOR data.

During speech synthesis, the usage time of the same digital data obtained by one analysis is set to a time inversely proportional to the playback speed of the VTR.

The synthesized sound synthesized in this way has a pitch.

Because the position of formants is the same as in normal spoken language,
The sound is similar to that of someone speaking quickly. This allows recording even when playing at a non-rated speed.

By using a device to which the above-mentioned DEMRCOR analysis and synthesis is applied, there is no need to keep an eye on the - side when performing fast forwarding, and it is possible to read the tape while listening only to the sound of the tape.

In addition, it becomes possible to listen to recorded airs, etc. in a short time.

Furthermore, by using the PARCOR analysis and synthesis method, audio reproduction over a wide range of reproduction speeds becomes easy.

Embodiments of the present invention will be described in detail below with reference to the drawings.

FIG. 1 shows the introduction of the present invention. In FIG. 1, 1
This is the conventional VTR part, excluding the speakers. 8W1 is a switch to select the playback speed.

1-1 is a reproduced sound output section of the VTR. Does SWZ output the playback sound as is, or does it use PAI? After performing COR analysis and pitch conversion, output a synthesized sound, or
This is a switch for selecting whether to output playback sound or not.

2 is the speaker, 3 is A for sampling the playback sound
D converter, 4 is a timer that provides a sampling clock, 5 is a CPU that performs PARCOR analysis processing, etc.
6 is the RO (RO) JC read-only memory that stores the CPU program, 1 is the arithmetic RAM (write/read memory), and 8 is the CPU (central processing unit).

PARC? which is obtained as a result of PARCOR analysis and pitch conversion by the CPU. R for storing OR data
The voice synthesizer synthesizes the voice using this data. 9 is a multiplexer that switches the address path and data bus between the CPU and the speech synthesizer.

10 is a T voice synthesizer. Reference numeral 11 denotes a low-pass filter and an amplifier that remove high frequency components contained in one synthesized sound. 8WJ is a switch that transmits one synthesized sound to the speaker, and is linked to gWJ.

That is, when it is selected that the VTR1 performs non-rated speed playback using the At the number of Nampu horses that can be used.

The audio is sampled by 1 interrupt processing via the programmer 1 and stored in RAMP. The $y puzzling frequency is set in the timer 4 by the CPUj so as to be proportional to the playback speed set by the switch sW1. Once a certain number of samples (for example, !50) have been sampled, the sampled data is subjected to a Hemming Quinod by software.

Extracted as Blame Data. The extracted frame unit data is calculated by autocorrelation coefficient calculation, pitch extraction, and calculation of one parameter by Durbin's solution.
An ARCOR analysis is performed.

The frame unit sampling data, which was data in the waveform region, was analyzed by i''ARcOR analysis to determine the pitch, am
, voiced/unvoiced distinction, and converted into parameter PARCOR data.

Stored in RAM8. VT set by switch 8 WJ to use the data stored in RAMJ
So that it is inversely proportional to the playback speed of R.

CPt7t sets the speech synthesizer 10. Speech synthesizer 10 by multiplexer 9.

RAMJ is accessed in parallel with CPU# by switching the address path and data bus.
The synthesized seido waveform is sent to the speaker 2 via a switch 8WJ that works in conjunction with a switch 8Wj, after passing through a low-pass filter and an amplifier 11# for removing harmonic components. , the audio is output.

In addition, an AD converter 1. for performing PARCOR analysis. Timer 4. C'PUJ, ROMg, IAMF, R
AMP, multi-breech 9.

As shown in Figure 2, Ll'1i11 dedicated to PARCOR analysis
, and a one-chip MPU (micro processing unit) 11.

In addition, although the above-mentioned example explained v'rm, it is not limited to this and can be implemented similarly to a universal voice reproducing device such as a cassette tape recorder.

Next, the effects of the present invention will be described below.

■ By performing PARCO* analysis, it is possible to preserve pitch and formant positions over a wide range.

■ By saving pitch and formant positions,
It is possible to output audio at the pitch of Budhen's spoken words,
And the above enables listening.

4111-Brief Descriptionll1tj4 is an overall configuration diagram showing an embodiment of the present invention;
FIG. 2 is an overall configuration diagram showing another embodiment of the present invention.

J-VTR, j-... speaker, S... mu-D converter, 4-j iv+, J-CPU, a-ROM.

1, #...1muM, 9...Multiplex, 1o.
...Speech synthesizer, 11...Low pass filter and amplification S.

Applicant's agent Patent attorney Suzu E Take Hikoga Figure 1

Claims (1)

[Claims] The audio output of the audio playback device that is playing at a non-rated speed is subjected to PMURCOR using a sampling wave number that is proportional to the playback speed.
Analyze and PARCO1! A variable speed one-voice playback device characterized by comprising means for obtaining data, and means for synthesizing speech by setting the usage time of the PAICOI data obtained by this means to a time inversely proportional to the playback speed. .