KR940005044B1 - Voice recognizing apparatus and voice recording method - Google Patents

Abstract

The device records a voice by one sound with a voice synthesizer (IC4) and memory (6), The device includes a voice record and recognition unit (1) which compares data of 64K SRAM (IC3) and recognizes a voice, an amplifier (5) which amplifies a voice signal of microphone (M), a memory (6) which stores voice data, a storage unit (7), a 1st micom (8) which inputs output data of a voice record and recognition unit (1), a displaying unit (3) which displays results, and a processor (9) which has a 1st micom (8) and displaying unit (3).

Description

Speech recognition device and voice registration method

1 is a block diagram of a conventional speech recognition device.

2 is a flowchart of operation in the voice registration mode according to FIG.

3 is a flowchart of operation in the voice recognition mode according to FIG.

4 is a block diagram of the voice recognition device of the present invention.

5 is a flowchart of operation in the voice registration mode according to FIG. 4;

6 is an operation flowchart in the voice recognition mode according to FIG.

* Explanation of symbols for main parts of the drawings

1: IC voice registration and recognition unit 3: IC indicator

5: IC amplifier 6: IC memory

7: IC storage unit 8: IC first microcomputer

9: IC processing unit, IC ₁ IC analog voice recognizer, IC ₂ IC digital voice recognizer, IC ₃ IC64K SRAM, IC ₄ IC voice synthesizer

The present invention relates to a voice recognition device and a voice registration method thereof, and in particular, to register a voice by one pronunciation, thereby simplifying the registration process.

Generally speaking, the specific speaker voice recognition IC is the core IC of the system for controlling the target device by voice, but at the current technology level, it does not recognize the voice of all ordinary people, and the control word set in the system is registered only by the voice registered by the voice. Speech can be recognized, which is called speech recognition.

The conventional voice recognition device is composed of a voice registration and recognition unit 1 and a processing unit 4 as shown in FIG.

The voice registration and recognition unit 1 receives an analog voice signal from the microphone M, converts the analog voice signal into a digital signal, extracts characteristic data of the voice from the digital signal, converts the voice data into predetermined data, and stores the predetermined data ( Analog voice recognizer (IC ₄ ; TC 8861) that generates a certain signal by comparing to the data already stored (in case of registration mode) or in comparison with the already stored data (in case of recognition mode). 64K SRAM (IC ₃ ) that stores the characteristic data of voice among the digital signals in the registration mode and the digital voice recognizer (IC ₂ ; TC 8864) in the registration mode and outputs the specific data stored in the recognition mode. And a processing unit 4 as a microcomputer 2 and an indicator 3 for displaying a voice registration and recognition result as a signal is transmitted to and received from the voice registration and recognition unit 1. Castle was.

The analog voice recognizer IC ₁ and the digital voice recognizer IC _{2 for} voice recognition are ICs having a specification that completes registration if the pronunciation is similar when 3 pronunciations are inputted and compared when voice registration is performed.

The voice registration operation of the conventional voice recognition device configured as described above will be described with reference to FIG.

After repeating the word twice through the microphone (M) it is determined whether the two pronunciations are similar (S ₁ -S ₃ ).

In addition, if the sound is pronounced as a third time and is similar to the pronunciation, the voice is converted into a digital signal through an analog voice recognizer and a digital voice recognizer (IC ₁ , IC ₂ ), and 64K SRAM (IC ₃ ) is characterized by the characteristic of the voice signal. Save the data to complete one word registration (S ₄ -S ₇ ).

And voice recognition operation will be described with reference to FIG.

The boehmite (M) When a pronounced voice registration and recognition unit 1, an analog speech recognizer (IC ₁₎ digital speech recognizer (IC ₂₎ and then to extract the characteristics of the analog audio data converted to digital data in the above extraction The feature data is compared with features of words registered in 64K SRAM (IC ₃ ) (S ₈ -S ₁₀ ).

Therefore, the 64K SRAM IC _{3 transmits} the number of the most similar word to the microcomputer 2 of the processing unit 4, and the microcomputer 2 displays the recognition result on the display 3 so that the recognition is completed (S ₁₁ -S). ₁₃ ).

However, such a prior art has to pronounce three words in order to register a word due to the characteristics of the analog voice recognizer (IC ₁ ) and the digital voice recognizer (IC ₄ ). There is a drawback to the hassle.

The present invention is to solve the above-mentioned shortcomings, and to provide a voice recognition device and a voice registration and recognition method thereof capable of registering a word with a single pronunciation using a conventional analog voice recognizer and a digital voice recognizer. There is this.

Hereinafter, an embodiment of the present invention for achieving such an object will be described in detail with reference to the accompanying drawings.

4 is a block diagram of the speech recognition apparatus of the present invention, wherein the speech recognition apparatus of the present invention comprises a storage unit 7, a voice registration and recognition unit 1, and a processing unit 9. Amplified voice recognition of the microphone (M) is converted into a digital signal, stored in the memory according to the control unit and reproduced by the amplifier (5), a voice synthesizer (IC ₄ ; M 6258), a memory 6 generating a voice signal, After receiving the signal from the storage unit 7, converting it into a digital signal, extracting characteristic data of the voice from the digital signal, converting it into a constant data, storing it in a predetermined portion (when registering mode) or comparing it with already stored data (recognition). In the case of a mode), the voice registration and recognition unit 1 that generates a predetermined signal includes an analog voice recognizer IC ₁ , a digital voice recognizer IC ₂ , and a 64K SRAM IC ₃ . The processing unit 9 which displays voice registration and recognition results as a result of sending and receiving a predetermined signal with (1) and generates a constant control signal to the storage unit 7 is composed of a first microcomputer 8 and an indicator 3. .

Referring to FIG. 5 of the present invention configured as described above, when a certain word is pronounced by the microphone M, the first microcomputer 8 of the first processing unit 9 is amplified by the amplification unit 5 of the storage unit 7. The voice synthesizer IC ₄ converts the digital signal into a digital signal, stores the same in the memory 6, and generates a voice signal with the first pronunciation (S ₂₀ -S ₂₂ ).

In addition, the voice synthesizer IC ₄ continuously generates a voice signal with the second pronunciation, and if the two pronunciations are similar, the voice signal is generated with the third pronunciation, resulting in three pronunciations (S ₂₃ -S ₂₇ ).

In the present invention, since the sounding recognizers IC ₁ and IC ₂ have to input three phonetic pronunciations for comparison, the same sound is generated three times, and the second and third pronunciations reproduced by the voice synthesizer IC ₄ are noise and the like. The first pronunciation is compared with the second and third pronunciation because it may differ from the first input pronunciation.

The analog voice recognizer IC ₁ of the voice register and recognizer ₁ receives the three pronunciations, and if similar, data of the characteristic part of the voice data in the 64D SRAM IC ₃ through the digital voice recognizer IC ₂ . Since the data is converted into schedule data and stored, registration of one word is completed (S ₂₈ -S ₃₃ ).

In addition, referring to FIG. 6, when a certain word is pronounced by the microphone M, the word is amplified by the amplifier 5 of the storage unit 7 and registered in the 64K SRAM IC _{3 of the} voice synthesizer IC ₂ . Compare the features with the features (S ₃₄ ) (S ₃₅ ).

Subsequently, features of voice data are extracted from the analog voice recognizer IC ₁ of the voice register and recognizer ₁ , and the features are compared with words registered in the 64K SRAM IC _{3 of} the digital voice recognizer IC ₂ . (S ₃₄ ) (S ₃₅ ).

After the comparison, the number of the most similar words is transmitted to the first microcomputer 8 of the processing unit 9, and the recognition is completed by displaying the recognition result on the display 3 (S _36- S ₃₈ ).

As described above, according to the present invention, voice registration can be performed by one pronunciation by using the voice synthesizer IC ₄ and the memory 6.

Therefore, while using a conventional analog voice recognizer and a digital voice recognizer, there is an effect that can simply register the voice to avoid the hassle of having to pronounce three times during voice registration and recognition.

Claims (2)

Amplifying the input voice signal, converting the digital signal into a digital signal, and storing the converted first voice signal and simultaneously converting the stored voice signal into the original voice signal and generating the second voice signal. Converting the signal into an original voice signal to generate a third pronunciation, and if all of the first to third pronunciations are similar, converting and storing the characteristic part of the voice signal into digital data and indicating that the voice is registered. A voice registration method of a voice recognition device. Amplify the voice signal of the microphone (M) with the amplifier (5), convert the amplified signal into a digital signal through the voice synthesizer (IC ₄ ), and stores and reproduces the converted signal in the memory (6) in accordance with the control signal The storage unit 7 which generates a voice signal, and digitally converts the signals of the storage unit 7 through the analog voice recognizer IC ₁ and the digital voice recognizer IC ₂ , and extracts the feature data of the voice from the digital signal. and by storing the 64K SRAM (IC _3), and the voice registration, voice recognition as compared to through the digital speech recognizer (IC ₂₎ for data to be applied is stored in the 64K SRAM (IC ₃₎ data from an analog speech recognizer (IC ₁₎ The first microcomputer 8 inputs the control signal to the voice registration and recognition unit 1 and the storage unit 7 during voice registration, and outputs data of the voice registration and recognition unit 1 during voice recognition. And a processor 9 for displaying the recognition result through the indicator 3. A voice recognition device, characterized in that.