JPH11153998A - Audio response equipment and its method, and computer readable memory - Google Patents

Abstract

PROBLEM TO BE SOLVED: To easily and efficiently determine the audio response result of a correct inter-precaution from an audio response result in which homophonous and different meaning words are included. SOLUTION: A speech recognition part 105 recognizes an inputted speech signal. When homo-phonous and different meaning words are included in the speech recognition result, a response is prepared based on information concerning the homophobous and meaning words while storing a read changing dictionary 110 in which words and information concerning the homophonous and different meaning words of words are made to correspond each other and the prepared response is subjected to a voice synthesis in a voice synthesis part 108 to be outputted to the input origin of the inputted voice signal. Then speech recognition result is determined based on the answer of the input origin with respect to the response.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a voice response apparatus and method for recognizing an input voice signal and making a response based on the recognition, and a computer readable memory.

[0002]

2. Description of the Related Art CTI (Computer Telep)
In a voice response system such as a honey integration system, voice from a telephone can be recognized, and voice from a telephone can be used as a voice command. Also, by incorporating an e-mail function into the voice response system, the e-mail can be read out by voice synthesis. In addition, as a method of designating a specific e-mail, Sub including a keyword by a certain voice
jet can be selected. At this time, if the input speech has a plurality of homonym candidates, there is the following method as a method for coping with this.

[0003] 1. 1. The voice response system uniquely determines any of the homonyms as a voice recognition result. 2. From the context before and after the speech recognition result, an appropriate speech recognition result is selected and determined from homonyms. The voice response system does not determine the voice recognition result from the homonym, but the user or the operator accesses the voice response system later and determines the voice recognition result from the homonym using a display or the like.

However, in the above-mentioned conventional voice response system, in case (1) above, confusion has arisen when it is necessary to differentiate homonyms. In the case of the above 2, a large amount of calculation and learning data are required, and in the case of proper nouns such as names, it becomes impossible to identify them. In the case of the above 3, the burden on the operator and the user increases, and it becomes difficult for an operator other than the user to determine the homonym candidate such as the name.

[0004] The present invention has been made in view of the above problems, and a voice response apparatus capable of easily and efficiently determining a correct voice recognition result from a voice recognition result including a homonymous word. And a method thereof, and a computer-readable memory.

[0005]

A voice response apparatus according to the present invention for achieving the above object has the following arrangement.
That is, a voice recognition device that recognizes an input voice signal and performs a response based on the recognition, the voice recognition device recognizing the input voice signal, a word, and a homonym of the word. A storage unit for storing a dictionary in which information is associated with the input speech signal, wherein a response based on information relating to the homonym is included in a result of the speech recognition performed by the speech recognition unit. An execution unit for executing the speech recognition based on a response from the input source with respect to the response.

[0006] Preferably, the execution means creates a question sentence for determining any of the homonyms as a speech recognition result based on information on the homonyms, Voice synthesis means for voice-synthesizing the question sentence created by the means, and outputs the question text voice-synthesized by the voice synthesis means to an input source of the input voice signal.

Preferably, the information relating to the homonym includes at least reading of the homonym, information indicating characteristics of the homonym, and information belonging to the homonym.

Preferably, the apparatus further comprises management means for managing transmission and reception of electronic mail via a network line.

Preferably, the voice recognition result of the input voice signal is an instruction to read out an e-mail including a predetermined character string from among the e-mails managed by the management means. If it contains homonyms,
The execution means executes a response based on the information regarding the homonym to the input source of the input voice signal, and the determination means determines the character string based on a response from the input source to the response. Then, an e-mail including the character string is read out to the input source.

Preferably, the voice recognition result of the input voice signal is an instruction to read out an e-mail from a predetermined sender in the e-mail managed by the management means. If the word is included,
The execution means executes a response based on the information on the homonym to the input source of the input voice signal, and the determination means determines the sender based on a response of the input source to the response. Then, the e-mail from the sender is read out to the input source.

A voice response method according to the present invention for achieving the above object has the following configuration. That is, a voice recognition method for recognizing an input voice signal and performing a response based on the recognition, the voice recognition step of recognizing the input voice signal, a word, and a homonym of the word. A storage step of storing a dictionary in which information is associated with the storage medium, and, if a homonym is included in the speech recognition result obtained by the speech recognition step, a response based on the information regarding the homonym is given to the input speech. An execution step of executing the signal on an input source; and a determination step of determining the speech recognition result based on a response from the input source to the response.

A computer readable memory according to the present invention for achieving the above object has the following configuration. That is,
A computer-readable memory storing a speech recognition program code for recognizing an input speech signal and performing a response based on the recognition, wherein a program code of a speech recognition step for recognizing the input speech signal, Words and
A program code for a storage step of storing in a storage medium a dictionary in which information relating to the homonym of the word is associated, and information on the homonym if the homonym is included in the speech recognition result of the speech recognition step. A program code of an execution step of executing a response based on the input source of the input voice signal, and a program code of a determination step of determining the voice recognition result based on a response of the input source to the response. Prepare.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of the present invention will be described below in detail with reference to the drawings. [First Embodiment] FIG. 1 is a block diagram showing a configuration of a voice response system according to a first embodiment of the present invention.

As shown in FIG. 1, the voice response system comprises:
Broadly speaking, the voice response device 103 and the voice response device 103
, A telephone 101 for making a telephone input. The telephone 101 of the first embodiment is connected to the voice response device 103 through the telephone line 102, but may be wireless. The telephone 101 is a telephone normally used.

Next, various components constituting the voice response apparatus 103 will be described.

Reference numeral 104 denotes a telephone management unit, and the telephone 101
, And a response corresponding to the telephone input is returned to the telephone 101. Reference numeral 105 denotes a voice recognition unit that recognizes a voice waveform input by telephone using a grammar dictionary 106. Reference numeral 106 denotes a grammar dictionary, which is a dictionary for speech recognition that describes words to be subjected to speech recognition and word connection information. Reference numeral 107 denotes a homonym extraction unit, which determines whether or not there is a homonym from the speech recognition result by the speech recognition unit 105.

Reference numeral 108 denotes a speech synthesizer which creates text data based on telephone input and synthesizes a speech waveform from the text data. In addition, a speech waveform is synthesized from the text data input from the read-replacement creating unit 109.
Reference numeral 109 denotes a transliteration creating unit, which is a homonym extracting unit 10
7, if there is a homonym in the speech recognition result, the dictionary is replaced with a question sentence in which the homonym part is read differently.
10 is created. Reference numeral 110 denotes a reading dictionary, which is a dictionary in which characters constituting words to be subjected to speech recognition and data pairs formed by reading the words are registered. An e-mail management unit 111 controls transmission and reception of e-mail via the network 112. Reference numeral 112 denotes a network, which connects to an external device and sends e-mails and FAs.
It is used for transmitting and receiving data such as X.

Next, processing executed by the voice response system according to the first embodiment will be described with reference to FIG.

FIG. 2 is a flowchart showing processing executed in the voice response system according to the first embodiment of the present invention.

Here, the processing executed by the voice response system will be described as an exchange between the telephone 101 and the voice response device 103.

First, in step S 201, the user uses the telephone set 101 to send a telephone line 1 to the voice response apparatus 103.
Make a call through 02. In step S202, the voice response device 103 stores the incoming call in the telephone management unit 104.
Accept as phone input. In step S203, the user utters a voice using the telephone 101. For example.
Here, "Shoulder" is uttered.

In step S204, the speech waveform indicated by the telephone input from the user is stored in the speech recognition unit 105 and the grammar dictionary 106.
Performs voice recognition. In step 205, the homonym extraction section 107 compares the speech recognition result with the grammar dictionary 106 to determine whether the homonym exists in the speech recognition result. If there is no homonym (step S
(NO at 205), the process proceeds to step S209. On the other hand, if a homonym exists (YES in step S205),
Proceed to step S206. Note that the determination in step S205 compares the speech recognition result with the grammar dictionary 106. If a plurality of homonyms are registered in the grammar dictionary 106 and the ambiguity cannot be removed only by the connection information, the speech recognition result is determined. It is determined that a homonym exists. For example, the speech recognition result is “Make a chopstick”, and “Chopsticks” and “Hashi” are registered in the grammar dictionary 106 as words corresponding to “Mr. "And" Buy a bridge "are both connections allowed by grammar,
"Chopsticks" and "bridge" are considered as homonyms.

If it is determined in step S205 that a homonym exists, then in step S206, the homonym is read, for example, as shown in FIG.
The replacement sentence creating unit 109 creates a replacement sentence in which the homonym is read differently. Specifically, readings that do not overlap each other are taken out from the reading dictionary 110 and a sentence (text data) of “Is this a (phrase containing a homonym) in (homophonetic reading) of (reading)?” Is created. . For example, in the case of using the replacement dictionary 110 shown in FIG. 3 with respect to the above example, "Do you want to use a chopstick at (cho)?" Do you want to use a chopstick? "
In step S207, the text data created by the replacement sentence creation unit 109 is converted into a speech waveform by the speech synthesis unit 108, and is shown to the user through the telephone 101.

In step S208, the user listens to the voice-reading device 103 for the replacement sentence indicated by voice, and uses the voice or the push sound of the telephone 101 to find out which homonym is uttered by telephone input. Choose whether you uttered it intentionally. In step S209, the homonym selected by the user is determined as a speech recognition result, and transferred to a higher-level application. As an example of a higher-level application, a keyword corresponding to a speech recognition result and an e-mail of a sender are read out. Alternatively, the voice recognition result is returned as an e-mail.

In step S210, it is determined whether or not the end is input by a telephone input from the user. If the process has not been completed (NO in step S210), the process returns to step S203 to accept another telephone input. On the other hand, if the processing has ended (YES in step S210), the processing proceeds to step S210.
In step 211, the telephone is disconnected by the telephone management unit 104.

As described above, according to the first embodiment, when a homonym exists in the speech recognition result in response to a telephone input from a user, the homonym is presented to the user with different readings and confirmed to the user. So you can
An accurate speech recognition result can be determined without imposing a burden on the user. [Second Embodiment] In the first embodiment, the replacement dictionary 110 as shown in FIG. 3 is used. However, a configuration may be employed in which feature information indicating features of words other than the reading as shown in FIG. 4 is also registered. In this case, the homonyms can be determined by presenting the user with feature information that does not overlap with each other. Further, such a configuration is also effective in disambiguating keyword designation of an e-mail.

Hereinafter, a specific example will be described.

User (telephone set 101): Calls "chopsticks" to the voice response apparatus 103.

Voice response device 103: When the e-mail of "opening bridge" and the e-mail of "sales network of chopsticks" are managed by the e-mail management unit 111, the telephone input Of "Meru yoyome", "Hashi" is regarded as a homonym. Then, non-overlapping readings of "bridge" and "chopsticks" or feature information are selected from the reading dictionary 110 shown in FIG. 4, and the reading sentences are shown to the user. In this case, the user should see the following sentence for the user: "Is it a (Hashi) at (Hashi)" for "(Fashion)"? Is shown.

User (telephone 101): Here, for example, if the content uttered by telephone input is intended to be "(Hashi) of (Fuji)", "Fujiji no Hoshi" is displayed. Enter by phone.

Voice response device 103: Homophonic word "bridge"
And read the e-mail of "Bridge opening work".

As described above, according to the second embodiment, since the feature information indicating information on homonyms other than the reading is registered in the replacement dictionary 110, the first embodiment is used.
In addition to the effect obtained in the above, it is possible to more flexibly confirm the homonymous words to the user. [Embodiment 3] In the replacement dictionary 110 of Embodiments 1 and 2, the pronunciation and characteristic information are managed in units of characters. However, as shown in FIG. 5, a characteristic for distinguishing homonyms in advance for each word as shown in FIG. Replacement dictionary 110 in which information is registered
May be used. When such a replacement dictionary 110 is used, it can be used to specify a sender of an e-mail.

The operation example will be described below.

User (telephone 101): Telephone input to voice response device 103, "Whether or not me is staying."

Voice response device 103: When e-mails from Kato and Kato are managed by the e-mail management unit 111, the voice response device 103 responds to the user as "Mr.

User (telephone 101): In response to this response, telephone input to the voice response device 103 is "Yome".

Voice response device 103: "Which one is it?"
And respond to the user.

User (telephone 101): Instructs the voice response device 103 to select from the replacement dictionary 110 feature information that does not overlap each other as words of "Kato" and "Kato", which are regarded as homonyms.

Voice response device 103: Based on an instruction from the user, non-overlapping feature information of the words "Kato" and "Kato" is read in the replacement dictionary 110 shown in FIG.
And shows the replacement sentence to the user. In this case, the user is presented with a replacement sentence of "Is it going to be (Katosetsu) of (XX Kensetsu)?"

User (telephone set 101): Here, for example, if the intention is "(Kent of XX Kensetsu)", telephone input of "Kent of XX Kensetsu".

Voice response device 103: Determines the homonym by "Kato" and reads out an e-mail from Mr. Kato.

As described above, according to the third embodiment, since the characteristic information indicating information on homonyms other than the pronunciation is registered in the replacement dictionary 110, the first embodiment is modified.
In addition to the effect obtained in the above, it is possible to more flexibly confirm the homonymous words to the user.

As described above, according to the first to third embodiments, the ambiguity of the homonym can be eliminated through the voice by answering the question of the voice response device 103 by the user. become. As a result, the user can eliminate the ambiguity of the homonym by using only the telephone 101 without using an operator or special equipment.

Even if the present invention is applied to a system composed of a plurality of devices (for example, a host computer, an interface device, a reader, a printer, etc.), a single device (for example, a copying machine, a facsimile machine) Etc.).

Another object of the present invention is to provide a storage medium storing a program code of software for realizing the functions of the above-described embodiments to a system or apparatus, and to provide a computer (or CPU) of the system or apparatus.
And MPU) read and execute the program code stored in the storage medium.

In this case, the program code itself read from the storage medium realizes the functions of the above-described embodiment, and the storage medium storing the program code constitutes the present invention.

As a storage medium for supplying the program code, for example, a floppy disk, hard disk, optical disk, magneto-optical disk, CD-ROM, CD
-R, a magnetic tape, a nonvolatile memory card, a ROM, or the like can be used.

When the computer executes the readout program code, not only the functions of the above-described embodiment are realized, but also an OS (Operating System) running on the computer based on the instruction of the program code. ) May perform some or all of the actual processing, and the processing may realize the functions of the above-described embodiments.

Further, after the program code read from the storage medium is written into a memory provided in a function expansion board inserted into the computer or a function expansion unit connected to the computer, based on the instruction of the program code, It goes without saying that the CPU included in the function expansion board or the function expansion unit performs part or all of the actual processing, and the processing realizes the functions of the above-described embodiments.

[0050]

As described above, according to the present invention,
It is possible to provide a speech recognition apparatus and method capable of easily and efficiently determining a correct speech recognition result from a speech recognition result including a homonym, and a computer-readable memory.

[0051]

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a voice response system according to a first embodiment of the present invention.

FIG. 2 is a flowchart illustrating a process executed by the voice response system according to the first embodiment of the present invention.

FIG. 3 is a diagram illustrating a configuration of a replacement dictionary according to the first embodiment of the present invention.

FIG. 4 is a diagram illustrating a configuration of a replacement dictionary according to Embodiment 2 of the present invention.

FIG. 5 is a diagram illustrating a configuration of a replacement dictionary according to a third embodiment of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 101 Telephone 102 Telephone line 103 Speech recognition device 104 Telephone management unit 105 Speech recognition unit 106 Grammar dictionary 107 Homonymous word extraction unit 108 Speech synthesis unit 109 Transliteration creation unit 110 Translation dictionary 111 E-mail management unit 112 Network

Claims (13)

[Claims] 1. A voice response apparatus for recognizing an input voice signal and performing a response based on the recognition, comprising: voice recognition means for recognizing the input voice signal; a word; A storage unit for storing a dictionary in which information relating to the homonym is stored; and a case where the voice recognition result by the voice recognition unit includes a homonym, a response based on the information regarding the homonym is input to the input voice. A voice response device comprising: an execution unit that executes a signal input source; and a determination unit that determines the voice recognition result based on a response from the input source to the response. 2. A creating means for creating a question sentence for determining any of the homonyms as a speech recognition result based on information on the homonym, Voice synthesis means for voice-synthesizing the obtained question text, and outputting a question text voice-synthesized by the voice synthesis means to an input source of the input voice signal. The voice response device according to claim 1. 3. The homonym according to claim 1, wherein the homonym information includes at least information of the homonym reading, information indicating characteristics of the homonym, and information belonging to the homonym. Voice response device. 4. The voice response apparatus according to claim 1, further comprising management means for managing transmission and reception of electronic mail via a network line. 5. The voice recognition result of the input voice signal is an instruction to read out an electronic mail including a predetermined character string from among the electronic mails managed by the management means, and the character string includes a homonymous word. Is included, the execution means executes a response based on the information on the homonym to the input source of the input voice signal, and the determination means, based on a response of the input source to the response, The voice response device according to claim 4, wherein the character string is determined, and an e-mail including the character string is read out to the input source. 6. The voice recognition result of the input voice signal is an instruction to read out an e-mail from a predetermined sender in the e-mail managed by the management means, and the sender includes a homonymous word. The execution means executes a response based on the information relating to the homonym to the input source of the input voice signal, and the determination means executes the response based on the response of the input source to the response. 5. The voice response apparatus according to claim 4, wherein the voice response device reads out the e-mail from the sender to the input source. 7. A voice response method for recognizing an input voice signal and performing a response based on the recognition, comprising: a voice recognition step of recognizing the input voice signal; A storage step of storing, in a storage medium, a dictionary in which information relating to the homonym is associated; and when the speech recognition result of the speech recognition step includes a homonym, a response based on the information relating to the homonym is input to the input. A voice response method, comprising: an execution step to be performed on an input source of the input voice signal; and a determination step of determining the voice recognition result based on a response from the input source to the response. 8. The execution step includes: creating a question sentence for determining any of the homonyms as a speech recognition result based on the information on the homonyms; A voice synthesis step of voice-synthesizing the obtained question text, and outputting a question text that is voice-synthesized in the voice synthesis step to an input source of the input voice signal. Voice response method as described. 9. The method according to claim 7, wherein the information on the homonym includes at least reading of the homonym, information indicating characteristics of the homonym, and information belonging to the homonym. Voice response method. 10. The voice response method according to claim 7, further comprising a management step of managing transmission / reception of an electronic mail via a network line. 11. The voice recognition result of the input voice signal is an instruction to read out an electronic mail including a predetermined character string from among the electronic mails managed in the managing step, and the character string includes a homonymous word. Is included, the execution step executes a response based on the information regarding the homonym to the input source of the input audio signal, the determination step is based on a response of the input source to the response, 11. The voice response method according to claim 10, wherein the character string is determined, and an electronic mail including the character string is read out to the input source. 12. The voice recognition result of the input voice signal is an instruction to read out an email from a predetermined sender in the email managed in the management step, and the sender includes a homonymous word. The execution means executes a response based on the information relating to the homonym to the input source of the input voice signal, and the determining step includes determining the sender based on a response of the input source to the response. 11. The voice response method according to claim 10, wherein the e-mail from the sender is read out to the input source. 13. A computer readable memory storing a voice response program code for recognizing an input voice signal and performing a response based on the recognition, wherein the voice recognition step recognizes the input voice signal. A program code of a storage step of storing a dictionary in which a word and information on a homonym of the word are associated with each other in a storage medium; and a speech recognition result in the speech recognition step includes the homonym. In the case, the speech recognition result is determined based on a program code of an execution step of executing a response based on the information regarding the homonym to the input source of the input voice signal, and a response of the input source to the response. Computer-readable memory comprising: