JPH0562998B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION (a) Field of Industrial Application The present invention relates to a voice recognition device in which the state of an identification target in the next confirmation process changes depending on the result of voice confirmation.

(B) Prior Art Generally speaking, the recognition rate of speech recognition devices decreases as the number of recognized words increases, as shown in Japanese Patent Application Laid-Open No. 59-219798. This is more important than before,
The recognition rate was improved by dividing the recognition words into several groups and limiting the next recognition target words according to the results of speech recognition.
For example, when controlling electrical appliances using voice recognition, if the voice ``light'' is first recognized, then
Next, only the voices of the three words "on", "off", or "cancel" can be recognized. Second
The figure shows an example of such state transition rule information, and each state (state 1, state 2, state 3,...)
Information on state transitions for recognition target words and recognition results is stored. For example, let's say you want to control a light, a cooler, and a timer attached to the cooler using voice recognition. The 15 words used are from the number ``ichi'' to ``kiyuu'', ``light'', ``cooler'', ``timer'', ``on'', ``off'', and ``trikeshi''. The x in the figure means that the word is not recognized, and each number indicates the next state number to which the word will transition when it is recognized.

It is assumed that the cooler is currently stopped, and this is set as state 1. In state 1, the only words to be recognized are "light" and "cooler." First, if the voice "light" is input and recognized, a transition is made to state 2. In state 2, any one of "on", "off", and "trigger" is recognized, and the state returns to state 1.
When "on" or "off" is recognized, a predetermined signal is output from the output section.
In this case, the state 1 is the same when the light is on and when the light is off, but if it is inconvenient for a signal indicating that the light is on to be output when the light is on, a different state may be used. Just define it.

On the other hand, if it is recognized as a "cooler" in state 1, the state transitions to state 3. Here, only "on" and "trikeshi" are to be recognized, and if "on" is recognized, the state further transitions to state 4 and the cooler starts. Once the cooler starts, you can set the timer, and the words to be recognized are "Light",
The three words are "cooler" and "timer."

If "timer" is recognized in state 4, the state transitions to state 7, where the words to be recognized are numbers from "ichi" to "kiyu", "off", and "torikeshi". When a number is recognized, a signal is output to set the time until the cooler stops, when "off" is recognized, a signal is output to cancel the timer setting, and when "trikeshi" is recognized, a signal is output to set the time until the cooler stops. In either case, the process returns to state 4.

Conventionally, in order to memorize and control such state transition rules, the speech recognition device is controlled by a host computer, and the next recognition target word is selected by the speech recognition device based on the recognition result output from the speech recognition device. A method of informing has been used. However, this method requires a host computer, which tends to make the system complex and large.

On the other hand, the system can be simplified by storing state transition rules in the speech recognition device and allowing the speech recognition device itself to monitor state transitions and determine recognition target words.

FIG. 3 is a block diagram showing an example of the configuration of such a system, in which 1 is a microphone, 2 is a feature extractor that extracts the features of the input voice, and 3 is the feature of the input voice obtained by the feature extractor 2. This is a pattern creation section that creates an input pattern using . 4 is a standard pattern storage unit that stores a plurality of standard patterns registered in advance; 5 is a pattern matching unit that compares the input pattern with the standard pattern in the standard pattern storage unit 4; and 6 is a pattern matching unit that stores a plurality of standard patterns registered in advance. This is an output section that outputs a predetermined signal to the outside in response to the signal. 7 is a state transition rule storage unit for storing state transition rules as shown in FIG. 2; 8 is a standard for selecting a standard pattern in the standard pattern storage unit 4 using the information in the state transition rule storage unit 7; A pattern specifying section 9 is a gate for selecting a standard pattern to be sent to the pattern matching section 5 in accordance with the output from the standard pattern specifying section 8.

However, in the past, the state transition rules were incorporated as part of the microprocessor program that controlled the speech recognition device, which resulted in the speech recognition device being specialized, and the speech recognition device program was changed depending on the purpose. There was an inconvenience that I had to do.

(c) Problems to be Solved by the Invention The present invention is intended to solve the above-mentioned inconveniences, and provides a speech recognition device that can change the internally stored state transition rules by simple means. The purpose is to provide

(d) Means for solving the problem A feature extraction unit that extracts features from input speech, and a feature extraction unit that identifies the input speech based on the extracted features and outputs a predetermined signal according to the result. It is composed of an identification unit that outputs an output, and a state transition rule storage unit that stores transition rules between a plurality of states regarding an identification target that transitions according to the identification result of the identification unit, and the state transition rule storage unit is It is characterized in that it has a structure that can be attached to and detached from the identification section.

Furthermore, the state transition rule storage section is a ROM (read-only) installed on the IC socket.
memory).

(E) Effect Since the present invention is configured as described above,
By determining the identification target according to the rules in the state transition rule storage unit, the identification unit can easily apply state transition rules according to the purpose to the speech recognition device by simply replacing the state transition rule storage unit, for example, the ROM. can be added.

(F) Embodiment FIG. 1 shows an embodiment of the present invention. In order to embody the embodiment shown in FIG. This figure was realized by replacing it with the previous figure, and parts having the same functions as those in FIG. 2 are given the same numbers. The audio input through the microphone 1 is amplified by the amplifier section 10, and the BPF
The signal is divided into frequency bands by a band pass filter (band pass filter) 11, digitized by an ADC (analog-to-digital converter) 12, and taken into a microprocessor 13. The microprocessor 13 uses the arithmetic circuit 14 to create a voice input pattern based on the voice information. At the time of voice registration, a specified number of input patterns are stored as standard patterns in the standard pattern storage section 4, and at the time of voice recognition, the microprocessor 13 stores the input patterns and each standard pattern stored in the standard pattern storage section 4. The comparison is made using the arithmetic circuit 14, and the result is outputted to the outside through the output section 6. At this time, the microprocessor 13 refers to the state transition rules stored in the state transition rule storage section 7 based on the previous speech recognition results and determines the recognition target word.

The difference between the device according to the embodiment of the present invention and the conventional device lies in the state transition rule storage section 7'. That is, the state transition rule storage unit 7' stores state transition rules.
ROM71 and IC socket 7 for installing it
It consists of 2 parts, and the ROM can be easily replaced.

In this way, the ROM 7 of the state transition rule storage unit 7
1 is hardware-separated from the program storage unit 15 and can be replaced with another ROM that stores a state transition rule table corresponding to the purpose of the device.
The uses of such a device can be greatly increased without modifying the program.

(G) Effects of the Invention As is clear from the above explanation, according to the present invention, state transition rules can be used with a simple structure without requiring a host computer, and the state transition rules can be easily changed. Because you can
It is possible to provide an excellent speech recognition device that has a high recognition rate and can be used in many applications.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of an embodiment of the speech recognition device of the present invention, and FIG. 2 is a schematic diagram of a specific example of state transition rules.
FIG. 3 is a block diagram of a speech recognition device using state transition rules. 1... Microphone, 4... Standard pattern storage unit, 7'... State transition rule storage unit, 13... Microprocessor, 14... Arithmetic unit, 71...
ROM, 72...socket.

Claims (1)

[Scope of Claims] 1. A feature extraction unit that extracts feature amounts from input speech, and a feature extraction unit that identifies the input speech from among a large number of recognition targets based on the extracted features and determines a predetermined amount according to the result. an identification unit that outputs a signal, and a state transition rule storage unit that stores transition rules between a plurality of states regarding an identification target that transitions according to the identification result of the identification unit, and the identification unit includes: A speech recognition device characterized in that it performs identification processing after determining an identification target according to the rules of the state transition rule storage unit connected in a detachable state. 2. The speech recognition device according to claim 1, wherein the state transition rule storage unit is comprised of a ROM (read only memory) detachably installed on an IC socket.