JPH01295553A - Voice dialer - Google Patents

Abstract

PURPOSE:To attain accurate revision of a voice data and additional registration of a destination by adopting the constitution such that a required data is retrieved more efficiently among registered data or a dial number already registered in a voice dialer is easily retrieved. CONSTITUTION:The user registers a standard pattern and a dial number corresponding to the standard pattern for voice recognition and reply at first. When the registration data is not called by a voice input, the user retrieves a registration data to a control section 9 from a keyboard 10 and gives a command to apply an output in voice sequentially. Thus, a control section 9 outputs the voice sequentially from the voice output section 3. Moreover, the control section 9 reads the dial number of the registration data in voice output from a number storage section 8 to apply display output on a display section 11. When the outputted registration data is an object data, the data is designated and when an erroneous data is designated, a command is given to call a preceding data or a data to be outputted next from the keyboard. The operation is repeated to retrieve an object data.

Description

[Detailed Description of the Invention] The present invention provides a voice dialing device, more specifically, a voice dialing device that recognizes a destination name input by voice from among a plurality of destination names registered in advance, and The present invention relates to a voice dialing device that performs automatic dialing based on results.

Subdivision 1 - Conventional dialing devices for telephones require the dialing number of the other party's telephone subscriber to be input by manual dialing when making a call. This type of dialing device places a heavy burden on the user, such as having to memorize the dial number of the other party when making a call or refer to the dial number of the other party in a memo or the like. Therefore, a system has been proposed in which a dial signal is automatically output by voice input instead of manual dial operation (Japanese Patent Laid-Open Publication No. 144157/1983).

According to this method, the user can memorize the dial number,
There is no need to operate dials while looking at notes etc. In this type of device, it is necessary for the user to register in advance standard patterns for voice recognition and voice responses and corresponding dial numbers. However, as the number of registered data increases, it may become difficult to recall the desired data because the user forgets what name they have registered, or the tone of their voice changes due to a cold or the like.

A method in which the registered data is sequentially output by voice, and the user specifies the data when the desired data is output by voice, and performs operations such as dialing on the specified data. is possible. However, with this method, the desired data may be missed due to the user being talked to by another person during the confirmation process, etc.
There is a drawback that if data is specified before the target data is output, the registered data must be output from the beginning again.

Additionally, dialing devices that use voice recognition generally
It has a search function and displays or responds to the relationship between word numbers of registered words and phone numbers. In practice, this search function is used when you do not know what name it was registered under, or when you do not know whether it is registered or not. However,
General search function.

The phone number is displayed by inputting a word number, and all the words are searched and displayed or responded to in order to find them. Therefore, in the worst case, it is necessary to search for all the words, making the search time-consuming. Furthermore, when clusters are used for the search, the search range can be narrowed down by inputting a cluster number, but it is necessary to remember which cluster the subject belongs to.

In addition, in the case of voice dialing devices that perform telephone dialing by voice, conventionally it was possible to connect to only one phone number for one name on the other party. If you have multiple phone numbers and call one of them, and the phone is in use, you must manually disconnect the line and call again using the same name, or use a different phone number. If you have to call back, or if you are from the same group, such as a company,
Or, if the first recipient's phone is busy, such as when the recipient's phone is nearby, you may have to wait a while and try again, or call another number.
In this way, you can only register one phone number for one name, so when you have something urgent to say to the other party, or when the other party has a long phone call, you can use the same phone number over and over again. It was a waste of time and effort because it had to be done.

FIG. 8 is a block diagram showing an example of a voice dial telephone, and the device shown in this figure includes a voice recognition unit 92 connected to a handset 91, and a voice recognition unit 92 connected to a handset 91, and It is equipped with an automatic dialing section 94 that controls the connection with the subscriber line 93,
If you input the name of the other party by voice from the handset 91.

The voice recognition unit 92 recognizes this name and selects one of the pre-registered dial numbers of the other party, and based on this selection result, the automatic dialing unit 94 dials the dial number of the other party. let

By the way, in such conventional voice dialing devices, the number of destination dial numbers that can be registered in advance is limited. Only the dialed number of the destination that has already been registered must be deleted.

In this case, I used the search function to look at all the registered destinations and intuitively selected those that I didn't plan to use in the future, and then deleted them, so I used the search function to look at all the registered destinations, and then intuitively selected those that I didn't plan on using in the future and deleted them. Sometimes I ended up deleting the .

In addition, in such conventional voice dialing devices, it is impossible to achieve a 100% voice recognition rate, and since the speaker's voice changes over time, the recognition rate also gradually changes over time. There is a tendency to decrease.

For this reason, already registered audio data is often updated or re-registered.

However, in this case, since it is not known which destination name has a poor recognition rate, there is a problem in that it is difficult at the user level to update or re-register accurate voice data.

The present invention was made in view of the above-mentioned circumstances.
In particular, to more efficiently search necessary data from registered data in a voice dialing device that automatically dials out based on voice recognition results.
Alternatively, to facilitate the search for already registered dial numbers of a voice dialing device;
Can store multiple phone numbers for one name,
If the other party's phone is busy, it will be detected that the call is in progress, the line will be automatically disconnected, and a call will be made to a different phone number with the same name. It is an object of the present invention to provide a voice dialing device that can automatically extract destinations whose voice recognition rate has decreased, and thereby accurately perform additional registration of destinations, update of voice data, etc. It was done for a purpose.

Sei Tsuba - In order to achieve the above-mentioned object, the present invention includes a feature extraction section that extracts the feature amount of the voice for a name input by voice, and a standard pattern for the feature amount extracted by the feature extraction section. a standard pattern storage section that stores dial numbers corresponding to the names inputted by voice; and a number storage section that stores dial numbers corresponding to names inputted by voice;
By performing pattern matching between the i quantity and the feature quantity stored in the standard pattern storage section, it is possible to determine which standard pattern the input voice corresponds to! ! , reads out a standard pattern corresponding to the recognition result of the pattern matching unit from the standard pattern storage unit or another standard pattern storage unit stored in advance, and synthesizes an audio signal from the standard pattern. A voice dialing device comprising: a voice output section that outputs a voice; and a transmission circuit that reads a dial number of a name corresponding to the recognition result by the pattern matching section from the number storage section and outputs a dial signal.

When searching for a pair of standard patterns and dial numbers (hereinafter referred to as registered data) stored in the standard pattern storage unit and number storage unit, all or some of the registered data stored in the storage unit is sequentially read out and voice output is performed. A means for outputting, a means for the user to confirm the voice output and specify the registered data outputted by the voice, and a means for performing operations such as dialing on the specified registered data,
If the specified registration data is incorrect, a means is provided to read out the registration data output before or after the specified registration data, so that the desired registration data can be efficiently searched. In addition, there is a feature extraction section that extracts the feature amount for the name input by voice, a standard pattern storage section that stores the feature amount extracted by the feature extraction section as a standard pattern, and a voice input. A number storage unit that stores the dial number corresponding to the name input by the user, and pattern matching between the feature quantity extracted by the feature quantity extraction unit during voice input and the feature quantity stored in the standard pattern memory unit stored in advance. a pattern matching unit that performs the process, a preliminary selection matching unit that limits the target Qi words in the dictionary using n-frame data at the beginning of the audio or data of a part of the audio, and a pattern matching unit that It is equipped with a main selection matching unit that obtains the final result, and when searching for a phone number, the target word is uttered, and only the target word after preliminary selection of that word is searched, or input by voice. a feature extraction unit that extracts the feature quantity of the voice for the name inputted, a standard pattern storage unit that stores the feature quantity extracted by the feature extraction unit as a standard pattern, and a standard pattern storage unit that stores the feature quantity extracted by the feature extraction unit as a standard pattern; A number storage unit stores a plurality of dial numbers, and pattern matching is performed between the feature quantity extracted by the feature extraction unit at the time of voice input and the feature quantity stored in the standard pattern storage unit stored in advance. A pattern matching unit recognizes which standard pattern corresponds to the standard pattern, and a standard pattern corresponding to the recognition result of the pattern matching unit is read from the standard pattern storage unit or another standard pattern storage unit stored in advance, and the standard pattern is read from the standard pattern storage unit or another standard pattern storage unit stored in advance. an audio output unit that synthesizes and outputs audio signals from the above; a transmission circuit that selects and reads the dialed number of the name based on the recognition result by the pattern matching unit from the two-number storage unit and outputs the dialed signal; In a voice dialing device that includes a state detection unit that detects a signal indicating the status of a telephone with a dialed number and a connection control unit that disconnects a line, it is detected that a telephone with a dialed number that is first dialed is busy. , the device automatically stops the call, automatically selects and reads the next dialed number from a number storage unit that stores a plurality of dialed numbers for one name, and outputs a dialing signal; a storage unit storing the name and dial number of the other party; a voice recognition unit that recognizes the other party by comparing the name of the other party input by voice with the name stored in the storage unit; a voice synthesis unit that synthesizes voice to answer back the destination name recognized by the voice recognition unit; a dialing unit that dials the dial number of the destination based on the recognition result of the voice recognition unit; The present invention is characterized by comprising a data collection section that collects recognition rate data of the section and/or dialing destination data of the dialing section.

Hereinafter, the present invention will be explained based on examples.

FIG. 1 is a block diagram for explaining one embodiment of the present invention. In the figure, 1 is a handset, 2 is a feature extraction section, 3 is an audio output section, 4 is a pattern matching section, and 5 is a first A standard pattern storage section, 6 a second standard pattern storage section, 7 a transmission circuit, 8 a number storage section, 9 a control section, 10 a keyboard, and 11 a display section. and a standard pattern for response and the dial number corresponding to the standard pattern. When registering, if you group the standard pattern and dial number, you can call up the registered data read when searching. This is convenient because it allows you to narrow down your search to the group to which the registered data belongs. Here, the explanation will be given assuming that the data is registered after being divided into groups. Now, when the user is unable to call up the registered data by voice input as described in the section on the prior art, the user can access the control section 9 from the keyboard 10.
The system searches the registered data and gives instructions to output the data sequentially by voice. The control unit 9 instructs the audio output unit 3 to sequentially output the registered data by voice,
The audio output unit sequentially outputs audio according to the instructions. or,
The control section 9 reads out the dial number of the registration data that is being outputted by voice from the number storage section 8 and outputs the dial number to the display section 11 for display. At this time, if the group to which the registered data to be called belongs is known, by specifying that group, the data to be searched is narrowed down to that group; otherwise, all registered data is searched.

Further, although instructions are given using the keyboard here, any method such as commands by voice input may be used. next,
When the registered data output by voice is the data you want, you specify that data, but if you specify the wrong data here, you can use the keyboard (or voice command, etc.) gives an instruction to call up the previous data output or the next data to be output. If the retrieved data is not the desired data, this operation is repeated to search for the desired data. When the desired data is called up, the user performs operations such as dialing the data.

Furthermore, the voice dialing device generally registers the name of the subject by voice, and dials the pre-registered telephone number of the subject based on the recognition result. The recognition result is notified to the user by display or response. However, the user may find that the number of registered subjects increases, or that the user forgets how to register subjects that are infrequently used. The search function is used in such cases. The search function informs the user by displaying or responding by providing a number for a certain subject.

Its usage generally involves providing the registration number or scanning the numbers in some order to find and dial the appropriate subject.

However, in these searches, the user has to specify the search range, and in the worst case, the user has to search all the items to find the desired subject, which requires a lot of effort.

The embodiment shown in Figure 2 attempts to address these problems by prompting the user to make some kind of utterance, and automatically limits the search range based on the intermediate results of the recognition process. Inside, 21 is a telephone, 22 is a feature extraction unit, 2
3 is a code string storage unit, 24 is an analog conversion unit, 25 is a speech synthesis output unit, 26 is a dictionary pattern, 27 is a preliminary selection unit,
28 is a main selection section, 29 is a result output section, and 30 is a number reference section.

Reference numeral 31 designates a transmission circuit section, 32 a search section, ■ a voice response section, ■ a voice recognition section, and ■ a dial transmission section. The voice input from the telephone 21 is analyzed by the feature extraction unit 22 into respective feature quantities for recognition and response. and sent to the speech recognition section ■. The voice response unit I stores the obtained characteristic code string, converts it into an analog signal as needed, and outputs the response as a response. Next, the operation of the speech recognition section (2) will be explained. The speech recognition unit in this embodiment has a preliminary selection unit, which performs preliminary selection using a certain recognition method using the beginning or part of voice data to limit target words to a certain range, as shown in FIG. Then, after the voice utterance is finished, a main selection comparison is performed on those selected in the preliminary selection using all the voice data, and the results are output. The method of comparison between the preliminary selection and the main selection comparison at this time does not matter. In this embodiment, the preliminary selection result is used to perform the search.

When searching, first prompt the user to speak.

When users try to search, they either forget the pA record name for the subject, or for some reason they don't recognize it. Since I don't know if it was registered as "Section Manager" or something else, I just use whatever name I can think of when searching, such as "Kuninaka XXJ".
Say this out loud.

If a preliminary selection is performed using the audio data, only similar items such as ``Yu○O'' whose beginning begins with ``Yu'' will remain if the beginning data is used for the preliminary selection.

By displaying or responding to the subject names in descending order of similarity, the user can search and find the desired subject line with less effort.

Generally, the preliminary selection section narrows down the total number of targets to about 1/3 to 115. Therefore, it is possible to automatically narrow down the search target to that range.

Furthermore, conventional voice dialing devices that dial FFI calls by voice have only been able to connect to one phone number for one name on the other party's side. For example, if there are multiple phone numbers for one person on the other end, and you call one of them, and the phone is in use and is in a busy state, you can manually disconnect the line.
You will have to call again using the same name or call a different number. Also, if you are in the same group at work, or have a phone number nearby, if the first caller's phone is busy, you may have to wait for a while and then call again, or call another number. It won't happen. In this way, for one name.

If you can only register one phone number, you will have to perform the same actions over and over again when you have something urgent to say to the other party, or when the other party has a long call. The effort and time was selfless.

The embodiment shown in FIG. 4 solves this problem.
It is possible to memorize multiple phone numbers for one name, and if the other party's phone is busy, it will detect that the call is in progress, automatically disconnect the line, and switch to a different phone number with the same name. I was supposed to make a phone call.

In FIG. 4, 41 is a transmitter, 42 is a receiver, 43 is a communication circuit, 44 is a feature extraction section, 45 is a standard pattern for synthesis, 46 is a speech synthesis section, 47 is a connection control section, 48 is a control section, 49 is a standard pattern for recognition, 50 is a pattern matching section, 51 is a state detection section, 52 is a keyboard, 53 is a timer, 54 is a number storage section, 55 is a transmission circuit, 56 is a line control section, 57 is a central office line, This embodiment includes a feature extracting section that extracts the feature amount of the voice for a name input by voice, a standard pattern storage section that stores the feature amount extracted by the feature extracting section as a standard pattern, A number storage section that stores a plurality of dial numbers for a name input by voice, and a feature amount extracted by the feature extraction section when a voice is inputted and a feature amount stored in the standard pattern storage section that is stored in advance. A pattern matching unit recognizes which standard pattern the input voice corresponds to by performing pattern matching, and the standard pattern corresponding to the recognition result of the pattern matching unit is stored in the standard pattern storage unit or another pre-stored standard. An audio output section that reads out a voice signal from the pattern storage section, synthesizes and outputs an audio signal based on the standard pattern, and a dialing section that selects and reads out the dialed number of the name based on the recognition result by the pattern matching section from the number storage section and outputs the dial signal. In a voice dialing system, the telephone having the first dialed number is equipped with a circuit, a state for detecting a signal informing the state of the telephone for the first dialed number, a detection unit, and a connection control unit for disconnecting the line. If it detects that a call is in progress, the system automatically stops the call, automatically selects the first dial number from the number storage unit that stores multiple dial numbers for one name, reads it out, and dials it. This is a voice dialing method characterized by outputting signals.

The embodiment shown in FIG. 4 will be described in detail below.

First, the feature amount of the voice input from the transmitter 41 is extracted by the feature extraction unit 44, and the feature amount is selected as a standard pattern for recognition or a standard pattern for 1 synthesis depending on the state of the keyboard 52, or for pattern matching. The control unit 48 determines whether to perform the following. When the standard pattern registration button on the keyboard 52 is pressed.

Speech synthesis section 4 generates a voice synthesized sound as guidance for registering pre-registered voice information that prompts input of registered voice.
6 for the speaker. Thereafter, the speaker's voice is extracted by the feature extraction unit 44, and the feature amount is registered as a standard pattern for recognition and a standard pattern for synthesis. At this time, if there are multiple dial numbers for this voice, they will be dialed in order of frequency of use or convenience for the person using the phone.

Register in the number storage section from the keyboard.

Next, the time of recognition will be explained. When the standard pattern registration button on the keyboard is not pressed, synthesized speech is output to the speaker as guidance for recognition of pre-registered voice information that prompts the speaker to say the other party's name. After that, the speaker's voice is extracted by the feature extractor. This feature amount is matched with a standard pattern for recognition stored in advance in a pattern matching section, and the one with the highest degree of similarity is selected as the recognition result. This result is sent to the control unit, and if multiple phone numbers are registered in advance using the keyboard, one of them is automatically read out from the number storage unit. For example, the method is to read out the data in order of frequency of use and make a call.The above recognition results are also sent to the speech synthesis section, and standard patterns for synthesis, which are registered in advance and associated with each other, are read out.
Output to the receiver as a synthesized voice to confirm the destination of the call to the speaker.

Next, when the status detection unit detects that the other party to the first dialed number is on the phone, the current call is automatically stopped, the connection control unit disconnects the line, and the next call is automatically selected. This operation of reading out the dialed number from the number storage section and making one call continues unless the line is manually disconnected, and the numbers are selected in order from the most frequently used numbers to the lowest numbers and then to the highest numbers. When multiple telephone numbers are registered for one name, the information in this number storage section can be rearranged depending on the frequency of use, etc.

Therefore, according to this embodiment, a user can memorize a plurality of dial numbers for one name of the other party, and when dialing that one telephone, the telephone is in use and is in a call state. This eliminates the need to manually disconnect the line and then call again with the same name, or to dial a different number with the same name. Also, if the first recipient's phone is in a busy state, such as when there are phones in the same group, such as a company, or near the other party, the first number will automatically be set to the most frequently used number. If so, you can disconnect the line, select and read out the primary dial number, and output a dial signal when you have something to tell the other party, such as an urgent call.

This eliminates the need to repeat the same actions over and over again when the other party is on a long phone call, and reduces the annoyance of making a call and not being able to connect to the other party.

FIG. 5 is a block diagram showing another embodiment of the voice dialing device according to the present invention. The voice dialing device shown in this figure includes a voice recognition section 61, a voice synthesis section 62, a data storage section 63, A ring part 64, a keyboard part 65,
When the name of the other party is input by voice from the handset 67, the other party corresponding to this name is automatically dialed and connected to the subscriber line 68. Also, at this time, data on the voice recognition rate and frequency of use of dialed numbers are calculated, and these data are stored in the data storage section 63.

The voice recognition unit 61 includes a voice recognition circuit, and extracts voice data from the voice input from the handset 67 and stores it in the data storage unit 63, and extracts voice data from the voice input from the handset 67. and voice data stored in the data storage section 63, and the designated destination is recognized based on the voice.

Further, the data storage unit 63 includes a plurality of descriptors 69 and a message data area 74 in which voice message data is stored, as shown in FIG. 6, and each descriptor 69 indicates a destination. A voice data area 70 where voice data is stored, a dial number area 71 where the dial number of the other party corresponding to the voice data is stored, and the frequency of use of the dial number stored in this dial number area 71 is written. a usage frequency description area 72;
A recognition rate description area 73 is provided in which the recognition rate of the voice data stored in the voice data area 70 is written.

Then, the contents of these audio data area 70 to recognition rate description area 73 are rewritten as necessary.

Further, the speech synthesis section 62 includes a speech synthesis circuit,
When a control signal is supplied from the dialing section 64, the message data area 74 is accessed, voice message data based on the control signal is extracted, a message voice is synthesized, and this is output from the handset 67. Further, when an answerback command and destination data are supplied from the voice recognition unit 61, each of the descriptors 69 is accessed, voice data based on the destination data is extracted, and the destination name is is synthesized into audio and output from the handset 67.

Further, the keyboard section 65 includes a numeric keypad 75, a registration key 76, a search key 77, a delete key 78, an update key 79, a voice dial key 80, and a cancel key 81, as shown in FIG. When each of these keys is pressed, a corresponding key code signal is generated and supplied to the dialing section 64.

Further, the display section 66 is constituted by a display device such as an LCD display, and when display data is supplied from the dialing section 64, it is displayed on the display device.

The dialing section 64 also has a registration mode, a recognition mode, a deletion mode, a search mode, and an update mode, including the registration key 76, voice dial key 80, deletion key 78, search key 77, and update key. When any one of 79 is pressed, the mode corresponding to the pressed key is entered and each part of the device is controlled.

In this case, when the registration mode is selected, the voice synthesis unit 62 is controlled to output a predetermined guidance voice (guidance announcement) from the handset 67, to input the dial number of the other party manually, and to input the dial number of the other party. Allows voice input of the next 0 people. In parallel with this operation, a vacant descriptor 69 in the data storage section 63 is selected, and the dialed number data of the other party and the voice data of the name of the other party are stored in this descriptor 69 in correspondence with each other. Also, the usage frequency '''OIt is written in the usage frequency description area 72 of this descriptor 69.
, and the recognition rate "0" is also stored in the recognition rate description area 73.

In addition, when the recognition mode is selected, the voice recognition unit 6
1 to perform a matching process between the name of the other party inputted by voice from the handset 67 and the voice data registered in the data storage section 63, thereby recognizing the name of the other party inputted by voice.

Thereafter, the voice synthesis unit 62 is controlled to synthesize the recognized party's name and output it from the handset 67 (
If a message indicating that the called party who answered back is correct is inputted by voice or entered manually, the dialing unit 64 receives the dialed number of this called party from the data storage unit 63. reading the data and dialing it to connect the handset 67 and the subscriber line 68;
Enable calls. Also, at this time, the dialing section 64
The value of the usage frequency data stored in the usage frequency description area 72 of the descriptor 69 in which the dial number data of this destination was stored, and the value of the recognition rate data stored in the recognition rate description area 73. Increment by the value "1'".

In addition, in the above-described operation, if a message indicating that the person who answered back is incorrect is input by voice or keystrokes.

The dialing section 64 controls the speech synthesis section 62 to output a message for inputting the name of the other party again. Thereafter, when the name of the other party is inputted by voice from the handset 67, the voice recognition section 61 is controlled to perform a matching process between the name of the other party and the voice data registered in the data storage section 63. To recognize a destination name input by voice. After this, the above-mentioned answerback process is performed.

After that, if a message indicating that the called party who answered back is correct is inputted by voice or entered manually, the dialing section 64 reads out the dialed number data of this called party from the data storage section 63. At the same time, dialing is performed to connect the handset 67 and the subscriber line 68 to enable one call. At this time, the dialer 64 increments by "1" the value of the frequency of use data stored in the frequency of use description area 72 of the descriptor 69 in which the dialed number data of this destination is stored. In this case, since the voice recognition was not correctly recognized in the first voice recognition operation, it was not stored in the recognition rate description area 73, so the value of the recognition rate data stored in the recognition rate description area 73 is not incremented.

In addition, when the deletion mode is selected, the speech synthesis unit 6
2 and makes a predetermined guidance announcement to confirm with the user whether or not he or she wishes to perform a search for deletion. If a message that is not searched is input by voice or entered manually, the dialing section 6
4 controls the voice synthesis unit 62 to make a predetermined guidance announcement and have the user input the name of the other party to be deleted, and also controls the voice recognition unit 61 to recognize this other party and input the data. The contents of the descriptor 69 corresponding to this other party are erased from the storage unit 63.

In addition, in this deletion mode, when search processing is specified, the dialing unit 64 sequentially reads out the usage frequency data in the usage frequency description area 72 of each descriptor 69 in the data storage unit 63, and Sort (sort processing) in descending order of usage frequency. After this, the dialing section 64.

The voice synthesizing section 62 is controlled to select the destination that is least frequently used, and the destination name is voice synthesized and output from the handset 67, and the frequency of use is displayed on the display section 66. and.

When the user presses the delete key 78, the dialing unit 64 deletes the descriptor 69 of the currently selected destination. Further, if the user presses a key to display the next candidate, the dialing section 64 selects the second least frequently used destination, outputs this destination by voice from the handset 67, and The usage frequency of the other party is displayed on the display unit 66. Thereafter, the other party is deleted in the same manner. Then, when the cancel key 81 is pressed, the dialing section 64 ends this deletion mode.

Further, when the update mode is selected, the dialing section 64 controls the voice synthesis section 62, makes a predetermined guidance announcement, and confirms with the user whether or not to search for an update. If a message that is not to be searched is input by voice or entered manually, the dialing unit 64 controls the voice synthesis unit 62 to make a predetermined guidance announcement and notify the user.
The user inputs the name of the destination to be updated, and also controls the voice recognition unit 61 to recognize this destination, and the voice data area 7o of the descriptor 69 corresponding to this destination.
Update the contents written in.

In addition, in this update mode, when search processing is specified, the dialing unit 64 uses the usage frequency data in the usage frequency description area 72 of each descriptor 69 in the data storage unit 63 and the usage frequency data in the recognition rate description area 73. While reading out the recognition rate data sequentially, based on each of these values and the following formula,
The rate of correct recognition (correct recognition rate) is calculated for each destination.

Correct recognition rate = recognition rate data / usage frequency data... (1)
Thereafter, the dialing unit 64 sorts (sorts) the numbers in descending order of correct recognition rate. After this, the dialing part 6
4 controls the speech synthesis unit 62 to select the destination with the lowest correct recognition rate, synthesizes the name of the destination and outputs it from the handset 67, and outputs the name of the destination from the handset 67.
The usage frequency is displayed on the display unit 66. Then, when the user presses the update key 79, the dialing unit 64 controls the voice synthesis unit 62 to output a predetermined message from the handset 67 by voice, and input the name of the other party by voice. and when this is entered,
The voice recognition unit 61 is controlled to update the contents of the voice data area 70 of the descriptor 69 corresponding to this destination.

Furthermore, if the user presses a key to display the next candidate, the dialing unit 64 selects the other party with the second lowest correct recognition rate. Thereafter, update processing for the other party is performed in the same manner. Then, when the cancel key 81 is pressed, the dialing section 64 ends this update mode.

In this way, in this embodiment, the search function allows you to automatically extract destinations that are used less frequently, destinations whose voice recognition rate has decreased, etc., so that you can register additional destinations, Audio data can be updated accurately.

Furthermore, in the embodiment described above, when the update mode is selected and the search process is specified, the recognition rate and frequency of use are displayed.
Even when the value of the correct recognition rate is small, update registration can be canceled when the frequency of use is low.

Effects As is clear from the above description, according to the present invention, registered data can be efficiently searched in a voice dialing device even when the registered data cannot be called up by voice. Furthermore, since the search is performed using the preliminary selection results uttered during the search, it becomes possible to search and find the desired subject with less effort. In addition, the user can memorize multiple dial numbers for one name on the other party's side, and when dialing that one phone, or when the phone is in use and on a call, you can manually This eliminates the need to disconnect the line and call again with the same name, or to call a different number with the same name. Also, if the first recipient's phone is in a busy state, such as when there are phones in the same group, such as a company, or near the other party, the first number will automatically be set to the most frequently used number. With this, you can disconnect the line, select and read out the primary dial number, and output a dial signal. This allows you to output a dial signal when you have something urgent to say to the other party, or when the other party is on a long phone call. This eliminates the need to perform the same actions over and over again, and alleviates the annoyance of making a call and not being able to connect to the other party.

In addition, it is possible to automatically extract destinations that are used infrequently or whose voice recognition rate has decreased, making it possible to accurately register additional destinations and update voice data. .

[Brief explanation of the drawing]

FIG. 1 is a block diagram for explaining one embodiment of the present invention, FIG. 2 is a block diagram for explaining another embodiment of the present invention, and FIG. 3 is a block diagram for explaining another embodiment of the present invention. 4 and 5 are block diagrams showing other embodiments of the voice dialing device according to the present invention, and FIG. 6 is a pattern diagram for explaining the operation of the embodiment shown in FIG. FIG. 7 is a schematic diagram showing an example of a memory map of a data storage section, FIG. 7 is a plan view showing an example of the display section shown in FIG. 5, and FIG. 8 is a block diagram showing an example of a conventional voice dialing device. 1...Handset, 2...Feature extraction unit, 3...
Audio output section, 4... Pattern matching section, 5, 6... Standard pattern storage section, 7... Transmission circuit, 8... Number storage section, 9... Control section, 10... Keyboard , 11・
...Display section, 21... Telephone device, 22... Feature extraction section, 23... Code string storage section, 24... Analog conversion section, 25... Speech synthesis output section, 26... Dictionary pattern, 27...Preliminary selection section. 28... Main selection section, 29... Result output section, 30...
- Number reference section, 31... Transmission circuit section, 32... Search section, 41... Transmitter, 42... Receiver, 43...
Call circuit, 44... Feature extraction section, 45... Standard pattern for synthesis, 46... Speech synthesis section, 47... Connection control section, 48... Control section. 49... Standard pattern for recognition, 50... Pattern matching section, 51... State detection section, 52... Keyboard,
53...Timer, 54...Number storage unit, 55...
Transmission circuit, 56... Line control unit, 57... Office line, 6
1... Speech recognition unit, 62... Speech synthesis unit, 63...
-Storage section (data storage section), 64...Dialing section, 65...Keyboard section, 66...Display section, 67.
...Handset, 68...Line. Section 1, Section 2, Section 3, Section 5, Section 6, Section 7, Section 8, Section 5, Case Indication, 1988, Patent Application No. 224951, 2, Title of Invention, Voice Dial Device 3, Person Who Makes Amendment, Case and Relationship Patent applicant address Nakamagome, Ota-ku, Tokyo], Shimome 3-6 Name (674) Ricoh Co., Ltd. Representative Hama 1) Hiro 4, Agent address 231 Furo, Naka-ku, Yokohama Town 1-2-1
Chuo Dai6 Kannai Building 1001 December 20, 1986

Claims (1)

[Scope of Claims] 1. A feature extraction unit that extracts features of a name input by voice, and a standard pattern storage unit that stores the features extracted by the feature extraction unit as a standard pattern. a number storage unit that stores a dial number corresponding to the name inputted by voice, and a pattern of features extracted by the feature extraction unit at the time of voice input and features stored in the standard pattern storage unit stored in advance. A pattern matching unit recognizes which standard pattern the input voice corresponds to by performing matching, and a standard pattern corresponding to the recognition result of the pattern matching unit is stored in the standard pattern storage unit or another standard pattern stored in advance. an audio output unit that reads out a voice signal from the storage unit, synthesizes an audio signal based on the standard pattern, and outputs the synthesized voice signal; and a caller that reads out the dial number of the name corresponding to the recognition result by the pattern matching unit from the number storage unit and outputs the dial signal. When searching for a pair of standard patterns and dial numbers (hereinafter referred to as registered data) stored in a standard pattern storage section and a number storage section in a voice dialing device equipped with a A means for sequentially reading out some registered data and outputting audio, and a means for the user to confirm the audio output,
A means for specifying the registered data outputted by voice, and when performing work such as dialing on the specified registered data, if the specified registered data is incorrect,
A voice dialing device characterized in that it is equipped with a means for reading out registration data output before specified registration data or registration data to be output after specified registration data, so that target registration data can be efficiently searched. 2. A feature extraction unit that extracts feature quantities for a name input by voice, and a standard pattern storage unit that stores the feature quantities extracted by the feature extraction unit as a standard pattern;
A number storage unit that stores the dial number corresponding to the name input by voice, and pattern matching between the feature quantity extracted by the feature extraction unit during voice input and the feature quantity stored in the standard pattern storage unit stored in advance. a pattern matching unit that performs the above operations, a preliminary selection matching unit that limits the target words in the dictionary using n-frame data at the beginning of the audio or data of a part of the audio, and a pattern matching unit that Voice dialing characterized in that it is equipped with a main selection verification unit that obtains a final result, and that when searching for a telephone number, a target word is uttered, and only the target word after preliminary selection of the word is searched. Device. 3. The voice dialing device according to claim 2, wherein the voice dialing device displays or responds as a search result in descending order of similarity at the time of preliminary selection. 4. A feature extractor that extracts features of a name input by voice, a standard pattern storage unit that stores the feature extracted by the feature extractor as a standard pattern, and a name input by voice. A number storage unit stores a plurality of dial numbers for the name to be used, and pattern matching is performed between the feature quantity extracted by the feature extraction unit at the time of voice input and the feature quantity stored in the standard pattern storage unit stored in advance. A pattern matching unit that recognizes which standard pattern the input voice corresponds to, and a standard pattern corresponding to the recognition result of the pattern matching unit from the standard pattern storage unit or another standard pattern storage unit stored in advance. an audio output unit that reads out, synthesizes and outputs an audio signal from the standard pattern, and a transmission circuit that selects and reads a dial number in the name corresponding to the recognition result by the pattern matching unit from the number storage unit and outputs the dial signal; In a voice dialing device that includes a state detection section that detects a signal indicating the state of the telephone number that has been dialed, and a connection control section that disconnects the line, the telephone number that has been dialed first is in use. If this is detected, the system automatically stops the call, automatically selects the next dial number from the number storage unit that stores multiple dial numbers for one name, reads it out, and outputs a dial signal. A voice dialing device characterized by: 5. a storage unit storing the name and dial number of the other party, and a voice recognition unit that recognizes the other party by comparing the name of the other party input by voice with the name stored in the storage unit;
a voice synthesis unit that synthesizes voice to answer back the destination name recognized by the voice recognition unit; a dialing unit that dials the dial number of the destination based on the recognition result of the voice recognition unit; 1. A voice dialing device comprising: a data collection section that collects recognition rate data of the section and/or dialing destination data of the dialing section.