JPH04238398A - Voice recognition device and voice dialing device - Google Patents

Abstract

PURPOSE:To decrease the frequency of key depression for the registration of an opposite subscriber number and to improve the operability by using voice recognition even for the registration. CONSTITUTION:A feature extraction part 6 extracts the feature quantity of the voice of an opposite person name which is inputted from a microphone 1 and stores it in a standard pattern storage part 8. Further, the dial number of the opposite person is inputted before or after the registration of the voice by using a numeric voice dictionary stored as an unspecific voice recognition dictionary in a fixed standard pattern storage part 14 and stored in a dial number storage part 12 while related with a standard pattern. A command for switching between a recognition/origination mode and a registration mode is also stored in the fixed standard pattern storage part 9 and this is recognized to switch the modes.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a voice recognition device, and more particularly to an improvement in voice registration of a control device (eg, voice dialing device) using voice recognition.

0002

2. Description of the Related Art A so-called voice dialing device is being put into practical use that can automatically send a dialed number simply by speaking the name or nickname of the other party into a microphone. These voice dialing devices have a built-in specific speaker voice recognition device, and have an added function of transmitting a dial number corresponding to the recognition result.

FIG. 2 shows an example of the configuration of a conventional voice dialing device. First, the voice dialing operation will be explained. Assume that the dialing device is in the recognize/call mode. It is assumed that this mode switching is performed by the control section 11 controlling the entire system according to a signal inputted from the key switch 3. A feature extraction section 6 extracts the feature amount of the voice such as the name of the other party inputted from the microphone 1 of the handset. The extracted feature amounts are stored in the standard pattern storage unit 8 in the similarity calculation unit 7.
The degree of similarity with each of a plurality of standard patterns stored in advance is calculated. The result selection unit 9 regards the standard pattern with the highest degree of similarity as the recognition result, and selects the standard pattern with the highest degree of similarity.
0 and is passed to the control unit 11. The voice response unit 10 notifies the speaker of the recognition result by voice through the speaker 2 or the like. The voice for confirming the results may be synthesized from standard patterns for recognition, or feature amounts exclusively for synthesis may be separately prepared and synthesized from this. In the latter case, when registering the standard pattern for recognition, the output of the microphone 1 may be input to the response section and stored there. On the other hand, the control unit 11 that has received the recognition result stores the dial number corresponding to the standard pattern of the recognition result in the dial number storage unit 11.
Select from 2. Thereafter, the selected dial number is transmitted from the dial transmission section 13 to the telephone line 5 through the line control section 4.

[0004] Next, the registration operation will be explained. It is assumed that the registration mode is set by the key operation described above. A feature extraction section 6 extracts the feature amount of the voice such as the name of the other party inputted from the microphone 1 of the handset. The extracted feature amounts are stored in the standard pattern storage unit 8.
is memorized. Furthermore, before and after registering this voice, the dial number of the other party is input using the key switch 3, and is stored in the dial number storage section 12 in association with the standard pattern described above. In this way, once you register a voice dialing device, you can simply call the other party by voice.
Since you will be able to dial out calls, there will be no need to memorize numerous numbers. Furthermore, it is possible to provide a telephone that is very easy to use even for people with visual or physical disabilities, since there is no need to press dial buttons. However, as explained above, registration requires inputting the dialed number of the other party by pressing a key switch. Although this is not very difficult for a healthy person, it is a great burden for a person with visual or physical disabilities, and the advantages of the voice dialing device cannot be utilized.

[0005]

[Objective] The present invention has been made in view of the above-mentioned problems, and provides a control device using voice recognition, for example,
The purpose of the voice dialing device is to reduce the number of key presses during registration and improve usability by using voice recognition when registering the destination number.

[0006]

[Structure] In order to achieve the above objects, the present invention provides (1)
a feature extraction unit that extracts features of input speech; a standard pattern storage unit that stores feature quantities of a plurality of extracted speech sounds as a standard pattern; and a feature quantity of the extracted speech and storage in the standard pattern storage unit. a similarity calculation unit that calculates the degree of similarity with a plurality of standard patterns that have been calculated, and a result selection unit that selects one or more standard patterns as a recognition result in order of the calculated degree of similarity. In the speech recognition device, a part of the standard pattern storage section is configured with a non-rewritable ROM, and the remaining part is configured with a rewritable RAM, or (2) features of the input voice are extracted. a feature extraction unit, a standard pattern storage unit that stores extracted feature quantities of a plurality of voices as standard patterns, and similarity between the extracted voice feature quantities and the plurality of standard patterns stored in the standard pattern storage unit; and a result selection unit that selects one or more standard patterns as a recognition result in descending order of the calculated similarity. A part of the standard pattern stored in the section may be registered in advance as a speech pattern for non-specific speaker recognition, and the remaining part may be used as a speech pattern for specific speaker recognition, or (3) input. a feature extraction unit that extracts features of the extracted voices, a standard pattern storage unit that stores the extracted feature quantities of the plurality of voices as a standard pattern, and a control data storage unit that stores equipment control data corresponding to the standard pattern. a similarity calculation unit that calculates the similarity between the extracted voice feature amount and a plurality of standard patterns stored in the standard pattern storage unit; Alternatively, in the speech recognition device, the speech recognition device includes a result selection unit that selects a plurality of standard patterns as recognition results, and a control data output unit that outputs control data stored in the control data storage unit based on the selection results. A part of the standard pattern stored in the standard pattern storage unit is registered in advance as a voice pattern for non-specific speaker recognition, and the remaining part is used as a voice pattern for specific speaker recognition, registering the control data corresponding to the standard pattern for specific speakers using the voice for unspecified speaker recognition, or (4) a feature extraction unit for extracting features of the input voice; a standard pattern storage unit that stores feature quantities of a plurality of voices as a standard pattern; a dial number storage unit that stores a dial number corresponding to the standard pattern; and a standard pattern storage unit that stores the extracted voice feature quantities and the standard pattern storage unit. a similarity calculation unit that calculates the degree of similarity with a plurality of standard patterns stored in the computer; a result selection unit that selects one or more standard patterns as a recognition result in order of the calculated degree of similarity; In the voice recognition dialing device, the voice recognition dialing device includes a dialing unit that outputs the dialed number stored in the dialed number storage unit based on the selection result, and a part of the standard pattern stored in the standard pattern storage unit is The voice pattern for specific speaker recognition is registered in advance, and the remaining part is used for the voice pattern for specific speaker recognition, and when registering the dial number corresponding to the standard pattern for the specific speaker. (5) a feature extraction unit that extracts the features of the input speech and a standard that stores the extracted features of the plurality of speech as a standard pattern; a pattern storage section;
a dial number storage unit that stores a dial number corresponding to the standard pattern; and a similarity calculation unit that calculates the similarity between the extracted voice feature and the plurality of standard patterns stored in the standard pattern storage unit. a result selection unit that selects one or more standard patterns as a recognition result in descending order of the calculated similarity, and outputs the dial number stored in the dial number storage unit based on the selection result. In a voice recognition dialing device, a part of the standard pattern stored in the standard pattern storage part is registered in advance as a standard pattern representing a numerical value or a control word, and the remaining part is registered as a standard pattern representing a number or a control word. Further, (6) The voice dialing device of (5) is characterized in that the standard pattern representing the numerical value or control word is stored in a non-rewritable ROM as a standard pattern for speaker-independent recognition. Hereinafter, the present invention will be explained based on examples.

FIG. 1 is a block diagram for explaining one embodiment of the present invention. A standard pattern storage section 8 that stores voice features as a standard pattern, a dial number storage section 12 that stores dial numbers corresponding to this standard pattern, and the extracted voice features and the standard pattern storage section 8. a similarity calculation unit 7 that calculates the similarity with a plurality of standard patterns stored in the computer; and a result selection unit 9 that selects one or more standard patterns as a recognition result in order of the calculated similarity. , and a dialing unit 13 that outputs the dialed number stored in the dialed number storage unit 12 based on the selection result. A part is registered in advance as a standard pattern representing a numerical value or a control word, and the remaining part is used for a voice pattern for the other party, and a dial corresponding to this standard pattern for the other party is configured. When registering a number, the standard pattern for numerical recognition is used. More preferably, in such a voice dialing device, the standard pattern representing the numerical value or control word is not rewritten. This feature is characterized in that it is stored in an impossible ROM as a standard pattern for speaker-independent recognition.

The configuration shown in FIG. 1 is different from the configuration shown in FIG. 2 in that the number of fixed standard pattern storage sections 14 is increased. The operation in the recognition/transmission mode is the same as that explained in the prior art shown in FIG. 2, so a description thereof will be omitted. The operation in the registration mode will be explained below. Voice input from microphone 1 of the handset, such as the name of the other party, is
The feature amount is extracted in the feature extractor 6. The extracted feature amount is stored in the standard pattern storage section 8. Furthermore, before and after registering this voice, the dial number of the other party is input using the numerical voice dictionary stored in the fixed standard pattern storage unit 14, which is an unspecified voice recognition dictionary.
The dial number storage unit 12 is connected to the standard pattern described above.
is memorized. Further, a command for switching between the recognition/transmission mode and the registration mode is also stored in the fixed standard pattern storage section 8, and mode switching is performed by recognizing this command. At the time of recognition, whether to use an unspecified dictionary or a specific standard pattern for the other party can be switched depending on the operation history of the device. Further, if there is no difference in the recognition operation between unspecified and specific, there is no need to switch. These operations will be illustrated with specific examples.

(1) When you say "registration" into the handset, "Touroku" in the fixed standard pattern 14 is designated as a recognition result, and the system enters registration mode. (2) Next, say the name of the other party. This is stored in the standard pattern storage section 8. (3) Next, say the dialed number of the other party one digit at a time. These are recognized from the unspecified dictionary (fixed standard pattern storage section 14), and the recognition/registration of the dial number is completed. (5) Normally, it is in recognition/calling mode and on standby. The parts directly related to the present invention have been described above. Note that the present invention does not specify or limit the voice recognition and response (synthesis) algorithms, feature amounts, etc. In addition, operations such as updating and deleting standard patterns and dial numbers are already known and will not be particularly explained. Further, since it can be easily inferred that the present invention can be applied to these operations such as updating and deletion, the explanation here will be omitted.

0010

[Effects] As is clear from the above explanation, according to the present invention, by using voice recognition to register the dialed number of the other party and to switch the mode, the number of key presses during registration can be reduced, and voice recognition has improved usability. Dialing equipment can be provided.

[Brief explanation of the drawing]

FIG. 1 is a block diagram for explaining an embodiment of a voice dialing device according to the present invention.

FIG. 2 is a diagram for explaining an example of a conventional voice dialing device.

[Explanation of symbols]

1. Microphone 2 Speaker 3 Key switch 4 Line control section 5 Line 6 Feature extraction section 7 Similarity calculation section 8 Standard pattern storage section 9 Result selection section 10 Voice response section 11 Control section 12 Dial number storage section 13　Dial transmission section 14 Fixed standard pattern storage section

Claims (6)

[Claims] 1. A feature extraction unit that extracts features of input speech; a standard pattern storage unit that stores feature quantities of a plurality of extracted speech sounds as a standard pattern; a similarity calculation unit that calculates the degree of similarity with a plurality of standard patterns stored in the pattern storage unit; and a result selection unit that selects one or more standard patterns as a recognition result in order of the calculated degree of similarity. 1. A speech recognition device comprising a standard pattern storage section, wherein a part of the standard pattern storage section is made up of a non-rewritable ROM, and the remaining part is made up of a rewritable RAM. 2. A feature extraction unit that extracts features of input speech; a standard pattern storage unit that stores feature quantities of a plurality of extracted speech sounds as a standard pattern; a similarity calculation unit that calculates the degree of similarity with a plurality of standard patterns stored in the pattern storage unit; and a result selection unit that selects one or more standard patterns as a recognition result in order of the calculated degree of similarity. In the speech recognition device, a part of the standard pattern stored in the standard pattern storage part is registered in advance as a speech pattern for non-specific speaker recognition, and the remaining part is registered as a speech pattern for specific speaker recognition. A speech recognition device characterized in that it is used for speech patterns. 3. A feature extraction unit that extracts features of input audio, a standard pattern storage unit that stores the extracted feature quantities of a plurality of audio as a standard pattern, and a device control data that corresponds to the standard pattern. a control data storage unit for storing; a similarity calculation unit for calculating the similarity between the extracted voice feature amount and a plurality of standard patterns stored in the standard pattern storage unit; A result selection section that selects one or more standard patterns as a recognition result in descending order of the highest standard pattern, and a control data output section that outputs control data stored in the control data storage section based on the selection result. In the speech recognition device, a part of the standard pattern stored in the standard pattern storage section is registered in advance as a speech pattern for non-specific speaker recognition, and the remaining part is registered as a speech pattern for specific speaker recognition. 1. A voice recognition control device, characterized in that the control data corresponding to the specific speaker standard pattern is registered using the voice for non-specific speaker recognition. 4. A feature extraction unit that extracts features of input audio, a standard pattern storage unit that stores the extracted features of a plurality of audio as a standard pattern, and stores a dial number corresponding to the standard pattern. a dial number storage unit; a similarity calculation unit that calculates the similarity between the extracted voice feature and a plurality of standard patterns stored in the standard pattern storage unit; A voice recognition dialer comprising: a result selection unit that sequentially selects one or more standard patterns as a recognition result; and a dialing unit that outputs a dial number stored in the dial number storage unit based on the selection result. In the ring device, a part of the standard pattern stored in the standard pattern storage section is registered in advance as a voice pattern for non-specific speaker recognition, and the remaining part is used as a voice pattern for specific speaker recognition. A voice recognition control device characterized in that the voice recognition for non-specific speaker recognition is used to register a dial number corresponding to the standard pattern for specific speakers. 5. A feature extraction unit that extracts features of input voices, a standard pattern storage unit that stores the extracted feature quantities of a plurality of voices as a standard pattern, and stores a dial number corresponding to the standard pattern. a dial number storage unit; a similarity calculation unit that calculates the similarity between the extracted voice feature and a plurality of standard patterns stored in the standard pattern storage unit; A voice recognition dialer comprising: a result selection unit that sequentially selects one or more standard patterns as a recognition result; and a dialing unit that outputs a dial number stored in the dial number storage unit based on the selection result. In the ring device, a part of the standard pattern stored in the standard pattern storage section is registered in advance as a standard pattern representing a numerical value or a control word, and the remaining part is configured to be used as a voice pattern for the other party. A voice dialing device characterized in that the standard pattern for recognizing numerical values or control words is used to register a dial number corresponding to the standard pattern for the other party. 6. The voice dialing device according to claim 5, wherein the standard pattern representing the numerical value or the control word is stored in a non-rewritable ROM as a standard pattern for speaker-independent recognition. voice dialing device.