JPH0438358B2 - - Google Patents

Description

[Detailed Description of the Invention] [Field of Application of the Invention] The present invention relates to a device for recognizing human speech.
In particular, the present invention relates to an improvement of a speech recognition device including a plurality of speech recognition means that recognize input speech information by comparing it with a pre-registered speech pattern.

[Background of the invention]

For example, as disclosed in Japanese Patent Application Laid-Open No. 51-28701, the current speech recognition accuracy is significantly degraded for unspecified speakers, so it is necessary to register speech patterns in advance that extract features for each speaker. The mainstream method is to compare this with the voice input and recognize it based on the matching of those features.

By the way, there is a strong demand to use a speech recognition device with multiple speakers, and in this case, one speech recognition device can be used to record the speech patterns for each word of multiple speakers (for example, 20 people) in advance. Let me remember it. Then, the words produced by the user (speaker) are compared with the stored speech pattern, and recognized as words that correspond to the speech pattern that is closest or within the expected error range.

However, remembering a large number of words from a large number of speakers requires a large memory capacity, and recognizing matching speech patterns from among them requires a large amount of storage capacity.
The disadvantage is that the recognition time is long.

For this reason, a method has been proposed in which each speaker owns a cassette memory in which each speaker's voice pattern is stored, and the cassette is set when using the speech recognition device.

According to this method, although the total memory capacity is the same, there is an advantage that recognition accuracy and recognition speed are improved by matching only the voice pattern of a specific speaker.

However, this is inconvenient for the user, as the cassette memory, etc., may be lost or stored.

[Purpose of the invention]

An object of the present invention is to provide a speech recognition device that does not impair recognition accuracy or recognition speed when multiple speech recognition devices are used by multiple speakers, and does not require removable memory. It is.

[Summary of the invention]

The present invention is characterized by a means for identifying a speaker using any one of a plurality of speech recognition means, a common auxiliary storage means for storing speech patterns for each of the plurality of speech recognition means, and a common auxiliary storage means for storing speech patterns for each of the plurality of speech recognition means. The common auxiliary storage means includes (1) a control means for reading and storing a voice pattern corresponding to a speaker identified from the auxiliary storage means into a voice pattern storage means of the corresponding voice recognition means; (2) A speech pattern for tasks (speaker change terms) and (2) a speech pattern for tasks (commands, task terms, etc.) are stored in advance, and immediately before using each speech recognition means, speaker identification is performed. The speech patterns for the speech recognition means are read out and stored in the speech pattern storage means in each speech recognition means.

In this state, any voice recognition means, for example, A
When Taro starts speaking, he utters a predetermined speaker change term that is encoded with a voice pattern for identifying the speaker, such as his name.

Accordingly, when the speaker is identified, the control means reads the working speech pattern corresponding to the identified speaker from the common auxiliary storage means and stores it in the speech pattern storage means in the corresponding speech recognition means. do.

Thereafter, voice recognition is performed by comparing only the voice pattern for work in one-on-one correspondence with the speaker and the voice input information of that speaker, and the necessary work is performed.

As a result, there is no need for a removable memory or the like, and the speed with which different speakers can simultaneously use a plurality of voice recognition means is not impaired.

[Embodiments of the invention]

Hereinafter, an embodiment in which the present invention is applied to a voice input storage device will be described in detail. In this embodiment, the device inputs and stores the worker's work and its results while providing voice guidance and answering. Furthermore, it is also possible to freely register voice patterns that should be performed for the first time or for a new speaker.

FIG. 1 shows the configuration of an embodiment of a voice information input storage device according to the present invention. In the figure, two voice recognition devices 1A and 1B are voice input microphones 6.
Amplifier 11A that amplifies the audio signals of A and 6B
and 11B, A/D converters 12A and 12B that convert audio signals into digital signals, audio pattern memories 14A and 14B that store audio patterns in advance, and audio that performs audio recognition by comparing input audio and audio patterns. It is composed of recognition control circuits 13A and 13B.

On the other hand, the voice output devices 2A and 2B include synthesized voice memories 22A and 22B that store voices for outputting voice, and voices that select and output the stored contents of the synthesized voice memories 22A and 22B according to the voice recognition result. Output control circuits 21A and 21
B. D/A converter 23 that converts the output signals of the audio output control circuits 21A and 21B into analog signals.
A and 23B, and amplifiers 24A and 24B which amplify the analog signal and make the answer back sound come out from the speakers (or earphones) 7A and 7B.

The auxiliary storage device 8 is commonly used by the plurality of speech recognition devices 1A and 1B, and stores speech patterns to be stored in the respective speech pattern memories 14A and 14B.

In addition, the control circuit 3 includes voice recognition devices 1A and 1B.
control the voice recognition control circuits 13A and 13B of the voice recognition control circuits 13A and 13B of the
A and 2B audio output control circuits 21A and 2
1B to output guidance and answer back sounds from the speakers 7A and 7B, to store the voice patterns in the voice pattern memories 14A and 14B of the voice recognition devices 1A and 1B in the auxiliary storage device 8, and vice versa. You can transfer the voice patterns in the storage device 8 to the voice recognition devices 1A and 1B or the voice pattern memories 14A and 14B of 1A or 1B, and display (or print out) the control status and voice recognition results on the display (or printer) 5. ) is a control computer. This control circuit 3 is controlled not only by voice but also by a keyboard 4.

Next, FIG. 2 shows an example of audio words used in an embodiment of the present invention.

The voice words consist of a speaker change term (speech pattern for speaker identification) for changing the speaker, a work term for performing a task, and a command word used for the task (speech pattern for the task).

First, to register the voice pattern, the speaker
A or 6B is used to sequentially read the audio words aloud, and the audio is sent to the audio pattern memory via the amplifier 11A or 11B, the A/D converter 12A or 12B, and the audio recognition control circuit 13A or 13B. 14A or 14B. The voice patterns stored in the voice pattern memory 14A or 14B are stored in the auxiliary storage device 8 with addresses assigned for each speaker.

Regarding the address allocation for storing voice words in the voice pattern memories 14A and 14B, command words and working words are stored at the same address common to all speakers, and speaker change words are stored at different addresses for each speaker. In the speaker change mode (at the start of use or when a change command is given), only the voice patterns of the speaker change terms of all speakers are stored in the voice pattern memory 14A or 14.
In the work mode after the speaker change is completed, the voice pattern of the command word and work term of one speaker identified in the change mode is transferred to the voice pattern memory 14A or 14B and audibly recorded. Enter data.

Next, an embodiment of audio information input according to the present invention will be described using FIG. 3.

In response to the voice guidance "What's your name?" from the speaker 7A, A-taro utters "A-taro" from the microphone 6A, which is stored in the speech pattern memory 14A by the speech recognition control circuit 13A of the speech recognition device 1A. The word "A-taro" that matches the input speech is searched out from among the speech words for speaker change, and its memory address or corresponding code is output to the control circuit 3.

The control circuit 3 receives the voice word code as data and displays it on the display 5, as well as issues a command to the voice output control circuit 21A to answer. Audio output control circuit 21
A outputs the voice data in the synthesized voice memory 22A according to the answer back command from the control circuit 3, and outputs the voice data in the synthesized voice memory 22A.
"A-Taro" is uttered from the speaker 7A via the A converter 23A and the amplifier 24A. Here, when Taro A utters and inputs "OK" from the microphone 6A, the voice recognition control circuit 13A of the voice recognition device 1A
The word "OK" that matches the input voice is selected from among the words for speaker change in the voice pattern memory 14A.
and outputs the address or code to the control circuit 3. The control circuit 3 thereby identifies that the speaker is A-Taro, reads out and stores A-Taro's working speech pattern stored in the auxiliary storage device 8 in the speech pattern memory 14A, and stores A-Taro's work speech pattern stored in the auxiliary storage device 8. While switching to work mode, turn on the audio output device 2A.
is controlled to emit voice guidance from the speaker 7A asking, "What's the work?"

When A-taro inputs ``in stock'' by voice, the speaker 7A returns a guidance saying ``What is the product number?'' as a result of voice recognition.For example, if he inputs ``1, 2, 3'' by voice, if it is recognized correctly, ``1'' will be returned. , 2, 3”
An answer backup is returned. Next, if you input "A" by voice in response to the guidance "Where is the parking lot?", the voice recognition will return "A" as an answer.

As a result of the above, A-Taro transfers his work voice pattern from A-Taro's voice from the auxiliary storage device 8 to the voice recognition device 1A, and then performs a highly accurate speech pattern by comparing it only with his own work voice pattern. In addition, using high-speed recognition, the data ``Product number 123 and warehouse A'' are entered.

When Taro A wants to finish his work, he inputs "change" through the microphone 6A, and the work mode is switched to the speaker change mode. That is, the control circuit 3 reads out the speaker change voice pattern from the auxiliary storage device 8 and stores it in the voice pattern memory 14A.

The above description has been made of the case where Taro A inputs voice from the microphone 6A, but the same applies to the case where Taro A inputs voice from the microphone 6B. speaker 7
In response to the voice guidance from B, “What is your name?”
When Taro A utters "A Taro" from the microphone 6B, the speech recognition control circuit 13B of the speech recognition device 1B selects a word that matches the input speech from among the speech words stored in the speech pattern memory 14B. It searches for "A-taro" and outputs its memory address or corresponding code to the control circuit 3. Under the control of the control circuit 3, "A-taro" is uttered from the speaker 7B via the amplifier 24B of the audio output device 2B. Here, when Taro A speaks and inputs "OK" from the microphone 6B, the voice recognition device 1
The voice recognition control circuit 13B of B searches the voice word "OK" that matches the input voice from among the words in the registered voice memory 14B and outputs its address or code to the control circuit 3. This results in
The control circuit 3 transfers the voice pattern for A-Taro's work stored in the auxiliary storage device 8 to the voice pattern memory 14B, sets it to A-Taro's work mode, and controls the voice output device 2B to output the speaker 7B. gives voice guidance asking, ``What's the work?'' Thereafter, the operation is exactly the same as when inputting audio from the microphone 6A.

This time, when B-taro inputs "B-taro" into the microphone 6A (or 6B), as a result of voice recognition, this time the voice pattern memory 14A (or 14B) is recorded.
B-taro's working voice pattern is entered from the auxiliary storage device 8, and B-taro can now input voice data.

Thereafter, in the same manner, the voice patterns for changing speakers and for working with multiple speakers stored in one set of auxiliary storage device 8 are derived to the plurality of voice recognition devices 1A and 1B to freely speak aloud. Ability to change personnel and enter data. The voice pattern may be registered from one voice recognition device and transferred to another voice recognition device via an auxiliary storage device.
Alternatively, the information may be registered from each voice recognition device.

Here, the auxiliary storage device 8 is an integrated circuit RAM or
It may be used as a ROM, or as a bubble cassette, cassette tape, floppy disk, etc. However, in order to freely register the voice pattern of a new speaker, a storage means other than ROM is used.

Reading and storing of voice patterns in the auxiliary storage device 8 and the registered voice memory 14A or 14B may be performed from the keyboard 4 in addition to voice input. Furthermore, by displaying the voice recognition results on the display 5, a plurality of speakers can alternately input voice information from a plurality of voice recognition devices even if the voice output devices 2A and 2B are omitted.

FIG. 4 shows the configuration of another embodiment of the audio information input device according to the present invention, and the same reference numerals as in FIG. 1 have the same functions. In the figure, radio mobile stations 30A and 30B transmit input audio from microphones 6A and 6B to antennas 33A and 33B.
transmitters 31A and 31B that emit radio waves from;
Receivers 32A and 3 receive radio waves from antennas 33A and 33B and generate voice guidance and answer back from speakers 7A and 7B.
2B. Radio fixed station 20
A and 20B are radio mobile stations 30A and 30
Receiver 21 receives the radio waves of B via antennas 23A and 23B and inputs them to voice recognition devices 1A and 1B of voice input/output devices 10A and 10B.
A and 21B, audio input/output devices 10A and 1
The output audio from the audio output devices 2A and 2B of 0B is transmitted to the receivers 32A and 32 of the radio mobile stations 30A and 30B via antennas 23A and 23B.
It is composed of transmitters 22A and 22B that emit radio waves to B. The voice pattern is registered by the speaker reading voice words one after another from the microphone 6A or 6B. Microphone 6A
Or the audio input from 6B is the radio mobile station 3
Radio waves are emitted from transmitters 31A or 31B of 0A and 30B via antennas 33A or 33B. This radio wave is transmitted by antenna 23A or 23
The received signal is received by the receivers 21A and 21B of the radio fixed station 20A or 20B via the radio terminal B, and is registered in the registered voice memory of the voice recognition device 1A or 1B.
The voice patterns registered in the registered voice memory are stored in the auxiliary storage device 8 with addresses assigned for each speaker. Further, the voice pattern stored in the auxiliary storage device 8 is transferred to the voice pattern memory of the voice recognition device 1A or 1B by operating the keyboard 4 or inputting voice to the voice recognition device 1A or 1B. When the speaker's voice data is input from the microphone 6A or 6B, it is received by the receiver 21A or 21B of the radio fixed station 20A or 20B through radio waves from the transmitter 31A or 31B of the radio mobile station 30A or 30B, and the voice is transmitted. Recognition device 1
Input to A or 1B. The answer box as a result of voice recognition is emitted from the voice output device 2A or 2B, and is emitted in the form of radio waves by the transmitter 22A or 22B. This radio wave is received by the receiver 32A or 32B and the speaker 7A or 7
Voiced by B. Speaker is microphone 6A or 6B
When inputting data by voice, an answer call or guidance is emitted from the speaker 7A or 7B, and the user inputs data by voice while opening the speaker.

In the embodiments described above, once a voice pattern is registered in one set of voice recognition devices, the voice pattern can be registered in another voice recognition device using the auxiliary storage device without uttering a voice.

In the above embodiment, the voice pattern for speaker change is also registered in the common auxiliary storage device 8, and only in the speaker change mode, each voice recognition device 1A, 1B
The data is stored in the voice pattern memories 14A and 14B in the internal memory. However, the voice pattern for speaker change is always stored in each voice pattern memory 14A,
14B may be stored.
In this case, at other addresses in each voice pattern memory,
Among the working voice patterns, patterns corresponding to the identified speaker are selectively stored.

In addition, although we have described a system that uses voice recognition means to identify speakers, it is also possible to enter a code for each speaker using a keyboard or any other input means; in this case, , the control circuit can easily identify the speaker.

〔Effect of the invention〕

According to the present invention, a plurality of speech recognition devices can be freely used by a plurality of speakers, and by identifying the speaker, the speech pattern of the corresponding speaker can be stored in the speech pattern storage means of the corresponding speech recognition device. Accordingly, it is possible to provide a speech recognition device with excellent recognition accuracy.

[Brief explanation of drawings]

FIG. 1 is a system configuration diagram showing the configuration of a voice information input device showing an embodiment of the present invention, and FIG.
FIG. 3 is a diagram showing an example of voice words used in the voice information input device shown in FIG.
FIG. 4 is a system configuration diagram showing the configuration of another voice information input device showing another embodiment of the present invention. 1A, 1B...Voice recognition means, 2A, 2B...
Audio output device, 3... Control means, 4... Keyboard, 5... Display device or printer, 6A, 6B...
...Microphone, 7A, 7B...Speaker, 8...Auxiliary storage means, 10A, 10B...Audio input/output device,
20A, 20B... Radio fixed station, 30A, 30
B... Radio mobile station, 14A, 14B... Voice pattern memory.

Claims (1)

[Claims] 1. A microphone, a voice pattern storage means for storing a voice pattern, and a voice recognition system that recognizes voice by comparing the voice pattern stored in the voice pattern storage means with the voice input from the microphone. and an auxiliary storage means for storing the speech pattern for each speaker, wherein a plurality of the speech recognition sections are provided, and the auxiliary storage means is connected to the plurality of speech recognition sections. The voice pattern stored in this auxiliary storage means is used as the voice pattern for speaker identification and the voice pattern for work, and the voice pattern for speaker identification stored in this auxiliary storage means is used as the voice pattern. means for storing in a storage means; speaker identification means for identifying a speaker by comparing the stored voice pattern for speaker identification with the voice input from the microphone; and a speaker identified by the identification. and means for writing a working voice pattern from the auxiliary storage means into the voice pattern storage means. 2. The speech recognition device according to claim 1, wherein the auxiliary storage means is a floppy disk.