JPH0442092A - Clock with speech recognition device and speech synthesis device - Google Patents

Abstract

PURPOSE:To enable easy recognition of present time even by a blind man by incorporating a voice input part, a picking-up part of characteristic, a recognition and judgement part, a speech regenerating part and a speech output part, into an inside of a clock. CONSTITUTION:Voice vocalized by a speaker (a clock owner) is subjected to sampling by a voice input part 1 consisting of a microphone, a sound range intensifying filter and an AD converter, into 8 KHz and 12 bits digital signals, for instance. At a picking-up part 2 of characteristic, voice signal converted into digital signal is fequency-converted by each frame of which time section is 16 ms long, and characteristic parameters thereof are picked up and then the characteristic parameters series of the vocalized work is obtained. The characteristic parameters series is compared with characteristic parameters series at frequency range of the speaker's specific word, which has been learned and stored in advance by a recognition and judgement part 3, and is recognized and judged there and then the word is sent to a central control part 4. At the central control part 4, a word to be output is decided based upon the sent word and the is sent to a speech regenerating part 8.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a clock, a speech recognition device, and a speech synthesis device.

[Conventional technology] Conventional technology includes analog clocks that display the current time by pointing the hour and minute hands at numbers on a dial, and digital clocks and clocks that display time numbers on a display such as a liquid crystal panel. Clocks with voice synthesizers were known that told the time by voice when a switch was pressed on the body.

[Problems and Objectives to be Solved by the Invention] However, with the conventional technology, there is a problem that visually impaired people cannot tell the time using an analog clock or a digital clock, and the conventional clock with a speech synthesizer cannot tell the time. However, you cannot tell the time without pressing the switch on the main body of the watch, so if you are far from the watch, you have to move to that location, which is very difficult for visually impaired people. This has the problem of being inconvenient.

Therefore, the present invention is intended to solve these problems, and its purpose is to enable anyone to know the time at a location away from the clock through voice input and voice output. .

[Means for Solving the Problems] A watch with a voice recognition device and a voice synthesis device includes a voice input unit that inputs voice and converts the voice signal into a digital signal, a signal received from the voice input unit, and features thereof. a feature extraction unit that extracts parameters; a recognition determination unit that receives a signal from the feature extraction unit and recognizes a word from the signal;
It has an audio decoding unit that composes the language to be output into a digital audio waveform signal, and an audio output unit that receives the signal from the audio decoding unit, converts the signal into analog, and outputs it as audio,
The watch is characterized in that the voice input section, the feature extraction section, the recognition determination section, the voice decoding section, and the voice output section are built into a watch.

[Example code] An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a system configuration diagram of a wristwatch equipped with a speech recognition device and a speech synthesis device according to the present invention.

As shown in FIG. 1, the wristwatch with a voice recognition device and voice synthesis device of the present invention includes a voice input section 1, a feature extraction section 2,
, recognition determination section 3, central control section 4, oscillation/frequency division circuit section 5,
Time counter part 6, date/day of the week counter part 7,
It is composed of an audio decoding section 8 and an audio output section 9.

The voice uttered by the speaker (that is, the owner of the watch of the present invention) is input to the voice input unit 1, which is composed of a microphone, a high-frequency emphasis filter, and an AD converter, at 8KHz, 12B.
It is sampled as a digital signal.

In the feature extraction unit 2, the audio signal converted into a digital signal is frequency-converted frame by frame, with a 16m5 section as one frame, and feature parameters in the frequency domain are extracted to obtain a feature parameter string of the uttered word. .

The extracted feature parameter string is compared in the recognition determination unit 3 with a feature parameter string in the frequency domain of a specific word of the speaker (i.e., the owner of the watch) that has been learned and stored in advance using the DP matching method. Then, it recognizes and determines what words have been uttered as sounds, and sends the words to the central control unit 4.

The central control unit 4 determines the language to be outputted based on the word, and sends the language to be outputted to the speech decoding unit 8.

The audio decoding unit 8 decodes the digital audio waveform corresponding to the language by searching and selecting the digital audio waveform stored in the audio waveform storage unit of the audio decoding unit 8, and outputs the digital signal as audio. Send to Department 9.

In the audio output section 9, which includes a DA converter, an amplifier, and a speaker, the sent digital signal is converted into an analog signal, amplified, and output as audio.

The timepiece of the present invention uses a time counter part 6 and a date/day counter part 7 for knowing the current time, date, and day of the week.
, are counted up by the clock signal generated by the oscillation/frequency dividing circuit section 5.

An example of the processing in the central control unit 4 will be explained along the flowchart shown in FIG.

When a voice with a volume equal to or higher than a threshold level, which is higher than daily noise, is input to the voice input unit 1, an interrupt signal is sent to the central control unit 4, and the microcomputer in the central control unit 4 normally handles the interrupt. It is in a waiting loop state (step 21).

When the speaker speaks the "time" or "date" at a volume that is above normal noise,
An interrupt is generated by speaking a voice such as
Exit the loop and proceed to step 22.

In step 22, the word uttered by the speaker is input, and it is determined whether the word is a time or date, and if so, it is determined that it is a command to the clock, and the process proceeds to step 23. If not, it is determined that it is noise and the process returns to step 21, a loop of waiting for an interrupt. This distinguishes between commands to the watch and normal conversation or noise.

If the speaker utters "time" in step 23, the process moves to time output processing in step 24, in which a language representing the current time is generated by referring to the time counter part 6, and a language expressing the current time is output as described above. Output the time.

If the speaker utters "date" in step 23, the process moves to date/day of the week high-power processing in step 25, in which a language representing the current date/day of the week is generated by referring to the date/day of the week counter part 7. , outputs the date and day of the week by voice as described above.

After the processing is completed, the process returns to the loop (step 21) that waits for an interrupt.

The above is an example of the sequence of human power and output by voice of the clock of the present invention.

[Effects of the Invention] As explained above, the speech recognition device and the speech synthesizer-equipped watch of the present invention have a speech input section, a feature extraction section, a recognition determination section, a speech decoding section, and a speech output section inside the clock. The built-in structure has the effect that even visually impaired people can easily tell the current time by voice from a distance from the watch.

[Brief explanation of the drawing]

FIG. 1 is a system configuration diagram of a watch with a voice recognition device and a voice synthesizer according to the present invention, and FIG. 2 is a flowchart showing the processing of the voice recognition device and a watch with a voice synthesizer according to the present invention.

Claims (1)

[Scope of Claims] (a) A watch with a voice recognition device and a voice synthesizer, (b) a voice input section that inputs voice and converts the voice signal into a digital signal; (c) a voice input section that inputs voice and converts the voice signal into a digital signal; a feature extractor that receives the signal and extracts its feature parameters; (d) a recognition determination unit that receives the signal from the feature extractor and recognizes words from the signal; and (e) a digital audio waveform signal representing the language to be output. (f) an audio output unit that receives the signal from the audio decoder, converts the signal into analog, and outputs it as audio; (g) the audio input unit, the feature extraction unit; A watch with a voice recognition device and a voice synthesizer, characterized in that the recognition determination unit, the voice decoding unit, and the voice output unit are built into the watch.