JPH04361299A - Voice recognition device - Google Patents

Abstract

PURPOSE:To obtain the voice recognition device which easily optimizes the amplification factor of an amplifying means. CONSTITUTION:When a voice is registered, a speaker is specified by a speaker specifying means, and the optimum amplification factor for each speaker is calculated by the voicing of a key word and stored in an amplification factor storage means; and when a voice is recognized, the speaker is specified by the speaker specifying means and then the amplification factor of the speaker is read out of the amplification factor storage means to set the amplification factor of the amplifying means. Consequently, the operation is facilitated.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a speech recognition apparatus, and more particularly to a method for setting an amplification factor of an amplification means for amplifying an input speech signal.

0002

[Prior Art] FIG. 4 shows, for example, Japanese Patent Application Laid-Open No. 58-125099
FIG. 1 is a block diagram showing a conventional speech recognition device disclosed in the above publication. In FIG. 4, 1 is a microphone that converts input audio into an audio signal, 2 is an amplification means that amplifies the audio signal, and 3 is a filter circuit 4 and a multiplexer 5, which will be described later. 4 is a filter circuit consisting of a plurality of BPFs (band pass filters) connected to the output side of the amplification means 2; 5 is a multiplexer that sequentially samples the output of the filter circuit 4 in a time-divisional manner; It is.

6 is an A/D converter that converts the analog output signal of the feature extraction means 3 into a digital signal; 7 is a microcomputer consisting of a processor, a clock generator, a memory, an input/output interface, etc.; A display device 9 displays operation guidance information, and 9 is an audio device that outputs operation guidance information by sound. The microcomputer 7 is connected to convert and set the amplification factor of the amplification means 2,
It is also connected to an A/D converter 6, a display 8, and an audio device 9.

FIG. 5 is an operation flow diagram showing the operation of a conventional speech recognition device when registering a voice. FIG. 6 is an operation flow diagram showing the operation of a conventional speech recognition device during speech recognition.

Next, the operation at the time of voice registration will be explained using the operation flow diagram shown in FIG. With conventional speech recognition devices,
Before performing voice registration, a keyword is input, the amplification factor of the amplification means 2 is set by inputting the keyword, and then the voice is registered.

The microcomputer 7 first requests voice input of a keyword through the display 8 or the audio device 9, and waits until the voice is input (S30). When the voice is input to the microphone 1, the microphone 1 converts the input voice into an audio signal, and the amplifying means 2 amplifies the audio signal output from the microphone 1. After that, when the amplified audio signal is output to the feature extraction means 3, the filter circuit 4 frequency-analyzes the audio signal using a plurality of BPFs, and the multiplexer 5 sequentially samples the output of the filter circuit 4 in a time-division manner. The /D converter 6 converts the output of the multiplexer 5 into a digital signal, and the microcomputer 7 creates an input pattern from the digital signal that is the output of the A/D converter 6 (S31).

Next, the microcomputer 7 detects the output audio signal level of the amplifying means 2 from the input pattern, and calculates the amplification factor of the amplifying means 2 so that the output audio signal level becomes a predetermined level (S32). As a calculation method for the amplification factor, there is a method in which the maximum value of the output audio signal level of the amplifying means 2 within the audio section is a predetermined level, and a method in which the average value of the output audio signal level of the amplifying means 2 within the audio section is a predetermined level. There is a way to get it to that level. Then, the microcomputer 7 generates a control signal for the amplifying means 2 according to its amplification factor (S33).

[0008] Thereafter, the microcomputer 7 requests the input of the registered voice through the display 8 or the audio device 9, and when the voice is input, creates an input pattern and
The input pattern is registered as a standard pattern, and when a predetermined number of standard patterns are registered, the registration ends (S34).

Next, the operation during speech recognition will be explained using the operation flow diagram shown in FIG. With conventional speech recognition devices,
Similar to voice registration, a keyword is input before executing voice recognition, the amplification factor of the amplification means 2 is set by inputting the keyword, and then the voice is recognized.

[0010] First, the microcomputer 7 requests the first voice input of the keyword via the display 8 or the audio device 9, and waits until the voice is input (S4).
0). When the voice is input to the microphone 1, the microphone 1 converts the input voice into a voice signal, and the amplification means 2
amplifies the audio signal output from the microphone 1. Thereafter, when the amplified audio signal is output to the feature extraction means 3, the filter circuit 4 frequency-analyzes the audio signal using a plurality of BPFs, and the multiplexer 5 sequentially samples the output of the filter circuit 4 in a time-division manner. The /D converter 6 converts the output of the multiplexer 5 into a digital signal, and the microcomputer 7 creates an input pattern from the digital signal that is the output of the A/D converter 6 (S41).

Next, the microcomputer 7 detects the output audio signal level of the amplifying means 2 from the input pattern, and calculates the amplification factor of the amplifying means 2 so that the output audio signal level becomes a predetermined level (S42). The microcomputer 7 then generates a control signal for the amplifying means 2 according to its amplification factor (S43).

[0012] Thereafter, the microcomputer 7 requests input of the voice to be recognized using the display 8 or the audio device 9. When the voice is input, it creates an input pattern and combines this input pattern with a plurality of registered standards. Performs matching with patterns and calculates similarity. If there is a standard pattern with this degree of similarity above the threshold, it is determined that the input voice is the same as the word voice registered as the standard pattern, and an identification signal corresponding to the recognition result is output from a predetermined output port ( S44).

[0013]

[Problems to be Solved by the Invention] Since the conventional speech recognition device is configured as described above, the amplification factor of the amplification means 2 can be optimized for each speaker by inputting a keyword before recognizing the speech. I needed to set it up.

The present invention has been made to solve the above-mentioned problems, and it is an object of the present invention to provide a speech recognition device that can easily set the amplification factor of the amplification means to an optimum value.

[0015]

[Means for Solving the Problems] The speech recognition device of the present invention specifies a speaker by a speaker specifying means at the time of voice registration, calculates an optimal amplification factor for each speaker by uttering a keyword, and storing the amplification factor in an amplification factor storage means;
When a speaker is specified by the speaker specifying means during speech recognition, the amplification factor for each speaker is read from the amplification factor storage means and the amplification factor of the amplification means is set.

[0016]

[Operation] The speech recognition apparatus according to the present invention automatically sets the optimum amplification factor for each speaker by simply specifying the speaker using the speaker specifying means during speech recognition.

[0017]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram of a speech recognition device showing an embodiment of the present invention. In FIG. 1, 1 to 9 are the same as those in the conventional example, so their explanation will be omitted. 10 is an amplification factor storage means for storing the amplification factor of the amplification means at the time of registration, and 11 is a speaker designation means for designating the speaker who speaks, and these are connected to the microcomputer 7. FIG. 2 is an operation flow diagram showing the operation at the time of voice registration in the embodiment of FIG. Figure 3
2 is an operation flow diagram showing the operation during voice recognition in the embodiment of FIG. 1. FIG.

The operation will be explained below. First, the operation at the time of voice registration will be explained using the operation flow diagram of FIG. The microcomputer 7 first requests designation of a speaker using the display 8 or the audio device 9, and waits until the speaker is designated (S1). The speaker designation is performed by the speaker designation means 11, and when the terminal 11a is connected to the terminal 11b side, the speaker ■ is designated, and the terminal 11a is connected to the terminal 11c.
When connected to the side, the speaker ■ is specified. When the speaker is designated by the speaker designation means 11, the microcomputer 7 determines from the output of the speaker designation means 11 whether the speaker ■ or the speaker ■ has been designated (S2).

If speaker (2) is selected, the microcomputer 7 sets a writing destination address for the standard pattern and amplification factor for speaker (2) (S3). If speaker (2) is selected, the microcomputer 7 sets the standard pattern and amplification factor writing destination address for speaker (2) (S4).

Next, the microcomputer 7 requests voice input of a keyword (S5), creates an input pattern (S6), and calculates the amplification factor of the amplification means 2 from the input pattern (S7). This is the same as the conventional example. Then, the microcomputer 7 stores the amplification factor in the amplification factor storage means 10 for the speaker ■ or for the speaker ■ according to the speaker designation result.
(S8).

Next, the microcomputer 7 generates a control signal for the amplification means 2 according to the amplification factor for the speaker designated by the speaker designation means 11 (S9). Thereafter, the microcomputer 7 requests the input of the registered voice through the display 8 or the audio device 9, and when the voice is input, it creates an input pattern, and also creates an input pattern for the speaker or speaker according to the speaker designation result. When a predetermined number of standard patterns are registered, the registration ends (S10).

Next, the operation during speech recognition will be explained using the operation flow diagram shown in FIG. microcomputer 7
First, a speaker specification request is made by the same operation as when registering the voice (S20), and a speaker is determined (S21). Speaker■
is specified, the microcomputer 7 reads out the amplification factor for speaker ■ from the amplification factor storage means 10 (S22).
. When speaker ■ is designated, the microcomputer 7 reads out the amplification factor for speaker ■ from the amplification factor storage means 10 (
S23). Then, the microcomputer 7 generates a control signal for the amplification means 2 according to the amplification factor for the speaker designated by the speaker designation means 11 (S24).

Thereafter, the microcomputer 7 requests the input of the voice to be recognized using the display 8 or the audio device 9. When the voice is input, it creates an input pattern and speaks according to this input pattern and the result of speaker designation. Matching is performed with a plurality of registered standard patterns for speaker ■ or speaker ■, and the degree of similarity is calculated. If there is a standard pattern with this degree of similarity above the threshold, it is determined that the input voice is the same as the word voice registered as the standard pattern, and an identification signal corresponding to the recognition result is output from a predetermined output port ( S25).

In the above embodiment, the number of speakers to be switched is two, but the number of speakers may be three or more.

In the above embodiment, the switching is possible for a plurality of speakers, but the number of speakers is limited to one, and the amplification factor stored at the time of voice registration is automatically read out at the time of voice recognition, and the amplification means The amplification factor may be set.

[0026]

As described above, according to the present invention, the speaker designation means is provided together with the amplification factor storage means, the speaker is designated by the speaker designation means at the time of voice registration, and the amplification factor of the amplification means is changed for each speaker. is set optimally and the amplification factor for each speaker is stored, and when a speaker is specified by the speaker specifying means during speech recognition, the amplification factor for each speaker stored in the amplification factor storage means is Since the structure is configured to read out and automatically set the amplification means to the optimum state, there is an effect that the operation becomes easy.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the configuration of a speech recognition device according to an embodiment of the present invention.

FIG. 2 is an operation flow diagram showing the operation at the time of voice registration in the above embodiment of the present invention.

FIG. 3 is an operation flow diagram showing the operation during speech recognition in the above-described embodiment of the present invention.

FIG. 4 is a block diagram showing the configuration of a conventional speech recognition device.

FIG. 5 is an operation flow diagram showing the operation of a conventional voice recognition device when registering a voice.

FIG. 6 is an operation flow diagram showing the operation of a conventional speech recognition device during speech recognition.

[Explanation of symbols]

1. Microphone 2 Amplification means 3 Feature extraction means 4 Filter circuit 5 Multiplexer 6 A/D converter 7. Microcomputer 8. Display 9. Sound equipment 10 Amplification factor storage means 11 Speaker designation means

Claims (1)

[Claims] 1. A voice input means for converting input voice into a voice signal, an amplification means for amplifying the voice signal, a speaker designation means for designating a speaker to utter the voice, and a voice designated by the speaker designation means. amplification factor calculation means for calculating an optimum amplification factor for each speaker; amplification factor storage means for storing the amplification factor calculated by the amplification factor calculation means; and an amplification factor for setting the amplification factor of the amplification means. and a setting means, when a speaker is specified by the speaker specifying means at the time of voice registration, an optimum amplification factor is calculated for each of the specified speakers from the input level of the predetermined word speech, and the amplification factor of the amplification means is optimized. and storing the amplification factor in the amplification factor storage means, and when a speaker is designated by the speaker designation means during speech recognition, the amplification factor is read out for each of the designated speakers from the amplification factor storage means. A speech recognition device characterized in that the amplification factor of the amplification means is set by the following.