JPS5960498A - Voice recognition equipment - 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 relates to speech recognition, and particularly to a speech recognition device that can be used by multiple people.

In recent years, greetings have attracted attention as a method of transmitting information, and are often applied to command input devices for electronic meters R, 14 in particular. There are a variety of vocal path and ij'ili devices, from unspecified speech systems that can recognize continuous utterances to inexpensive ones that limit speakers and words and can only handle discrete utterances. The pseudonyms are starting to dissolve. For specific speakers 4
The basic principle of a voice recognition device is to compare a pre-registered voice pattern with the voice input during use, and output the words that are considered to be the most similar in the form of a code. ing. Therefore, when using a neck %9 recognition device, two items are important: "registration" and "processing of comparison results."

With a speech recognition device for a specific speaker, if a speaker fails, it is necessary to redo the "registration", and when multiple people use the same device in turn, this semi-uniform work takes time. The problem arises that it takes a lot of time. Also, the host's electronic meter: j
The Ondo patterns of multiple people are registered in an auxiliary storage device such as a floppy disk by connecting the t-phone to a 4-voice recognition device for specific speakers, and the user can specify one of them with a key etc. Although it is conceivable to select the selected pattern and load it as a comparison pattern in the speech recognition device for a specific speaker, this method does not eliminate the need for extra operations for selection and the occurrence of erroneous operations. Therefore, there is a strong need for a method that can utilize an inexpensive speech recognition device for specific speakers to operate as if it were a speech recognition device for non-specific speakers, but until now there was no effective device. .

The present invention has been proposed in view of the above circumstances, and an object of the present invention is to provide a speech recognition device that can be used by a plurality of people using an inexpensive speech recognition device for a specific speaker.

The speech recognition device according to the present invention is a speech recognition device used by multiple people, in which a speech recognition unit for a specific speaker and a group of all patterns of words used for each individual are prepared for each person. It is composed of a group in which each individual's key word pattern is inputted by each individual's voice, and a total of n machine parts each having a memory for storing comparison registered patterns.
First, load the above key word pattern as a comparison pattern for the voice recognition unit in Section 1, and identify the user from the comparison result of the key word input by the user's first voice at the start of work. It is characterized in that the registered pattern is reloaded as a comparison pattern in the speech recognition section 1, and thereafter all words for personal use can be recognized.

An embodiment of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram showing a schematic configuration of an embodiment of the present invention, and FIG. 2 is a flowchart showing a processing procedure of the apparatus shown in FIG.

In FIG. 1, 10 is the host's electronic total 31! Machine, 11
is a voice recognition device for a specific speaker, 12 is an input/output device of Ichiko g[n=xo, such as a character display, 13 is a microphone as a voice input source, and 20 is a computer (hereinafter referred to as CPU).
), 21 is the memory, 22 is the key skewer metallurgy input by each person's neck μ, and all are Rfile, 23 to 27
28 is a character display interface; 29 is a voice recognition device interface; 30
is a CPU, 31 is a memory, 32 is a comparison pattern storage section,
33 is an interface for the host Nariko computer, 34 is a voice recognition central circuit, and 40.41 is an internal bus of each X machine system. Here, the registered patterns 22 to 27 may be placed not in the main memory but in an auxiliary memory such as a floppy disk.

Further, the key word in 22 may be a name, a code word, etc.

To explain the operation of the above embodiment of the present invention, the processing of the electronic meter S 1o in FIG. 1 is performed as shown in the flowchart of FIG. 2. That is, when the device is turned on, the electronic computer 10 transfers the key word registration pattern 22 stored in the memory 21 to the comparison pattern storage section 3 in the memory 31 in the speech recognition device 11 for a specific speaker.
2 via interfaces 29 and 33, respectively. In this state, the electronic meter H-Walo waits for the user to utter the key word of the condolence voice. When the user inputs the entire T-word into the microphone 13, which is the voice input source, it is encoded by the voice recognition central circuit 34, and this result is compared with the contents of the comparison pattern 32, and the results of the comparison are sent to the interface respectively. 33 and 29 to the host electronic meter nmxo.

When the host (Senke n Kai 10) receives this matching result, it determines whether it is from the royal palace or not (the user's input voice matches one of the comparison patterns), and identifies the discrepancies. If it is, it returns to the state of waiting for voice input.If it is strict,
The user is identified from the input key word, the corresponding registered patterns are sequentially outputted from the memory 21, and the respective registered patterns are interfaced to the comparison pattern storage section 32 in the memory 3 in the speech recognition device 11 for specific speakers. Transfer via faces 29 and 33. At this time, there is no previously loaded key word (fM pattern). From then on, the voice manually input by this user is compared with this personal registered pattern.

In other words, the process begins in accordance with the target system.

In the flowchart in Figure 2, when identifying users based on key words, it is possible to adopt a method of repeating the process multiple times to prevent misrecognition, or a method of preparing key words twice and checking them. It is possible.

Since the present invention is configured as described above, a key word pattern for identifying the user is initially included as a comparison pattern of the voice TG device for a specific speaker. Since the above-mentioned comparison pattern is automatically converted into a user's personal pattern, it is possible for the user to conveniently construct a speech recognition device for non-specific speakers. In addition, an excellent effect can be obtained in that an inexpensive microcomputer can be used as the host electronic computer.

In short, according to Fukōmei, in the head recognition device used by Karitetsujin, a specific speaker Kawane P recognition part and a group of all the patterns of the single g = used for each individual are used for the number of people. , a group in which key word patterns of each individual are input and registered by each individual's voice, and a total of n machine parts having a memory for storing comparison registered patterns including a group in which key word patterns of each individual are input and registered by each individual's voice. Load key word patterns and use the user's first key word pattern when starting work.
The user is identified from the comparison result of the key words entered by voice, and the registered pattern for the user is reloaded as the comparison pattern for the voice gX1 section, so that words for personal convenience can be recognized from now on. The present invention is extremely useful for business development because it provides a voice recognition device that can be used by multiple people using an inexpensive voice recognition device for a specific speaker.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing a schematic configuration of an embodiment of the present invention, and FIG. 2 is a flowchart showing a processing procedure of the apparatus shown in FIG. DESCRIPTION OF SYMBOLS 10... Electronic computer, 11... Speech recognition device for specific speakers, 12... Character display, 13...
Microphone, 20...CPU, 21...Memory, 22.
・Key words, 23-27 ・Useful words, 28゜29
...Interface, 30...CPU,,9Z...
・Memory, 32... Comparison pattern storage section, 33...
Interface, 34... Head recognition central circuit, 40° 41... Internal bus. Applicant Sub-Agent Patent Attorney Takehiko Suzue 6 Figure 2-626-

Claims (1)

[Claims] In a speech recognition device used by multiple people, a speech recognition unit for a specific speaker and a group of all word patterns for each individual are prepared for the number of people, and key word patterns for each individual are prepared for each individual. It consists of a total machine part having a memory for storing comparison registration patterns including a group input and registered by the individual's back door, and first loads the key word pattern as a comparison pattern of the sound P recognition oribe. , identify the user from the comparison result of the key word input by the user in the first voice at the start of work, and reload the registered pattern for the user as a comparison pattern for the voice recognition unit,
A voice-m device characterized in that it is capable of performing 111+1a of words for personal use thereafter.