JP2003271183A - Device, method and program for preparing voice recognition dictionary, device and system for recognizing voice, portable terminal device and program recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To prepare a voice recognition dictionary capable of recognizing a term even with incorrect reading. <P>SOLUTION: A first analysis dictionary storing part 5 stores a first analysis dictionary registered with the writing, reading, etc., of daily used terms. A second analysis dictionary storing part 6 stores a second analysis dictionary registered with the writing, reading, etc., of special terms. A text analyzing part 1 uses both the analysis dictionaries to perform a morphological analysis, and a reading attaching part 2 records a pair of a term and reading of the second analysis dictionary and a pair of a different candidate and a reading candidate in a correspondence table about registered terms of the second analysis dictionary in attaching reading to an input morpheme. A voice recognition dictionary preparing part 3 prepares a voice recognition dictionary on the basis of the contents of the correspondence table. As a result, the voice recognition dictionary is prepared on the basis of the association of a recognized term with phoneme writing of the second analysis dictionary and with a different phoneme writing candidate. Then, the use of the voice recognition dictionary can recognize the reading 'kyoshu' as 'kyoshu' regardless of pronouncing the reading 'kyosyu' as 'kyoowari'. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a voice recognition dictionary capable of correctly recognizing an utterance of a difficult-to-read word, and more particularly to a voice recognition dictionary creating apparatus and a voice recognition dictionary creating method, a voice recognition apparatus using the voice recognition dictionary, and a voice The present invention relates to a mobile terminal device equipped with a recognition device, a voice recognition system using the mobile terminal device, a voice recognition dictionary creating program, and a program recording medium recording the voice recognition dictionary creating program.

[0002]

2. Description of the Related Art In a voice recognition technique, an input voice can be recognized only by a reading method registered in advance in a vocabulary storage unit. Therefore, the user needs to know in advance what words can be recognized. In that case, if the vocabulary registered in the vocabulary storage unit is a small vocabulary, the registered vocabulary can be remembered to some extent. However, when the vocabulary registered in the vocabulary storage unit is a large vocabulary, it becomes difficult to remember the registered vocabulary.

As a method for solving such a problem, there are the following methods. (1) A method of displaying the vocabulary for voice recognition on the display means (for example, Japanese Patent Laid-Open No. 7-319383) (2) The vocabulary for voice recognition is dynamically changed according to the progress of the dialogue, and the voice recognition is performed. The target vocabulary is always a small vocabulary
(For example, Japanese Patent Laid-Open No. 6-332493) (3) By easily changing or adding a vocabulary to the vocabulary storage unit, it is possible to recognize a reading that is not a pre-registered reading (for example, Japanese Laid-Open Patent Publication No. (211893 publication)

[0004]

However, the conventional method for solving the problem when the vocabulary registered in the vocabulary storage unit is a large vocabulary has the following problems. That is, in the case of the method of displaying the speech recognition target vocabulary on the display means, the number of vocabularies that can be displayed on the display means is limited. In addition, when a place name or the like is displayed in Kanji, the user may not be able to read it correctly because the user does not know how to read it. In order to avoid this, if all place names are displayed in Hiragana notation, it will not be possible to correspond to the Kanji address written in a name list or a postcard. Furthermore, in order to avoid this, when both the kanji and hiragana notations are displayed together, only a small vocabulary can be displayed due to the limitation of the display area.

Therefore, the method of displaying the voice recognition target vocabulary on the display means is not sufficient to inform the user of the voice recognition target vocabulary including the reading.

Further, in the case of a method of dynamically changing the speech recognition target vocabulary according to the progress of the dialogue so that the speech recognition target vocabulary is always a small vocabulary, the speech recognition target vocabulary is dynamically changed. Therefore, the number of words to be recognized in each scene is small. However, in the end, it is necessary to remember all dynamically recognized recognition target vocabulary. Further, although the number of vocabulary to be recognized in each scene is small, the vocabulary is not recognized unless it is uttered according to the reading as stored in the vocabulary storage unit.

In particular, there are many difficult-to-read words in the case of a place name, and even if an attempt is made to retrieve a destination by utterance using a map search display device or the like, it may not be possible to use it because it is not known what to read. Or, I uttered it because I thought it was right,
It may not be recognized because it is different from the pronunciation read in the vocabulary storage.

As a specific example, for example, "Kyoto" in Nara is read as "Kyobate", but users who do not know often say "Kyoshu." However, in that case, since it is not recognized, the destination “Kyoto” cannot be searched and the map is not displayed. Even if the reading of the whole place name is not known, it may be found in units of one character, for example. For example, you cannot read "Kyoto", but you can read "Kyo" as "today" and "Kyu" as "shu" or "end." Therefore, the user predicts that a pair of "Kyo" and "Kyo", "End" and "Shu", or "End" is registered in the vocabulary storage unit, and then "Kyoto" is changed to "Kyoto". If you can recognize the voice by uttering "Shu" or "Kyosuru," it will be a solution when you cannot read "Kyoto." However, there is no guarantee that such a method will always work.

Similarly, the "new mouth" in Kashihara City says "Ninokuchi", but a user who does not know may say "Shinkuchi". If "Shinchi" is not registered in the vocabulary storage unit, it will not be recognized. Therefore, a user who thinks that "new mouth" is read as "shinchi" cannot display the map of "new mouth" forever.

By the way, in the case of the above example, by the method of changing or adding a vocabulary to the vocabulary storage unit,
This can be solved by additionally registering the difficult-to-read word and the reading that is likely to be misread in the vocabulary storage unit.

However, in any of the above-mentioned conventional methods, it is necessary for the user to know or be able to predict the recognition target vocabulary registered in the vocabulary storage unit. Therefore, in the above-mentioned conventional case, for example, when using words that are updated daily such as a TV program name and a music title name for voice recognition, the TV program name and the music title name are previously stored in the vocabulary storage unit. Cannot be registered, nor can it be predicted, and no voice recognition is possible.

As can be seen from the above examples, it is desirable that the voice is correctly recognized even when a word that is difficult to read correctly is uttered or the user does not know the correct reading.

Therefore, an object of the present invention is to provide a method in which the user does not know the correct reading of the recognition target word, or
Even if the user does not know the registered recognition target word, a voice recognition dictionary creating apparatus and a voice recognition dictionary creating method capable of creating a voice recognition dictionary capable of recognizing an input voice, and the above voice recognition dictionary are used. Voice recognition device, mobile terminal device equipped with this voice recognition device, voice recognition system using this mobile terminal device, voice recognition dictionary creation program,
Another object of the present invention is to provide a program recording medium which records the voice recognition dictionary creating program.

[0014]

In order to achieve the above object, a first aspect of the invention is to analyze an input text by a text analysis means, give a reading to the constituent words analyzed by the reading giving means, and perform voice recognition. In the voice recognition dictionary creating device for creating a voice recognition dictionary based on the analysis result and the reading addition result by the dictionary creating means, and storing the created voice recognition dictionary in the voice recognition dictionary storage means, the text analyzing means A first analysis dictionary storing means for storing a first analysis dictionary composed of information including vocabulary notation and reading, and a dictionary referred to when the text is analyzed by the text analysis means. A second analysis dictionary, which is composed of information including notations and readings of vocabulary not stored in the first analysis dictionary storage means, is stored. In addition to the second analysis dictionary storage means, the reading imparting means, when the text analysis result by the text analysis means includes a vocabulary obtained by referring to the second analysis dictionary, Regarding the vocabulary, in addition to the reading obtained by referring to the second analysis dictionary, other reading candidates are also added.

According to the above configuration, when the reading analysis means includes the vocabulary obtained by referring to the second analysis dictionary in the text analysis result by the text analysis means, the vocabulary is extracted as described above. In addition to the reading obtained by referring to the second analysis dictionary, other reading candidates are also added. Therefore, in the speech recognition dictionary created based on the analysis result and the reading addition result, regarding the vocabulary based on the second analysis dictionary, the phoneme notation of the reading obtained by referring to the second analysis dictionary and Dictionary information based on phoneme notation of other reading candidates is registered.

That is, for example, by registering the obfuscated word "Kyoto" and its correct reading "Kyobate" in the second analysis dictionary, the vocabulary "Kyoto" and its readings "Kyobate", "Kyobate" A voice recognition dictionary is created in which dictionary information based on "Shu" and "Kyo / End" is stored. Therefore, by performing voice recognition using this voice recognition dictionary, the target vocabulary “Kyoto” can be obtained as a recognition result without being rejected even if the user “utters” “Kyoushu” is mistakenly uttered.

In one embodiment, in the voice recognition dictionary creating device of the first invention, the contents of the second analysis dictionary stored in the second analysis dictionary storage means are acquired from the third dictionary storage means. The dictionary acquisition means is provided.

According to this embodiment, the contents of the second analysis dictionary are acquired from the third dictionary storage means by the dictionary acquisition means. Therefore, the newly appearing vocabulary is additionally registered in the second analysis dictionary storage means by providing the third dictionary storage means in which the new vocabulary is registered by the vocabulary information provider. Furthermore, even if the user does not know the recognition target vocabulary registered in the second analysis dictionary storage means, it becomes possible to create a voice recognition dictionary capable of recognizing the input voice.

A second aspect of the invention is to input the input voice,
In a voice recognition device for recognizing by performing matching with a vocabulary registered in a voice recognition dictionary, the voice recognition dictionary is created by the voice recognition dictionary creating device of the first invention. It is characterized by being.

According to the above configuration, with respect to the vocabulary registered in the second analysis dictionary, dictionary information based on the phoneme notation of reading obtained by referring to the second analysis dictionary and the phonetic notation of other reading candidates. Voice recognition is performed using the voice recognition dictionary in which is registered. Therefore, for example, the second
The obfuscated word "Kyoto" as an example of vocabulary registered in the analysis dictionary
Even if you accidentally say "Kyoushu", you will not be rejected and the target vocabulary "Kyoto" will be recognized as the recognition result.
Is obtained.

A speech recognition apparatus of a third invention is equipped with the speech recognition dictionary creating apparatus of the first invention, and the inputted speech is registered in the speech recognition dictionary storing means in the speech recognition dictionary creating apparatus. It is characterized in that the collation means performs collation with the vocabulary that has been recognized.

According to the above configuration, with respect to the vocabulary registered in the second analysis dictionary, dictionary information based on the phoneme description of the reading obtained by referring to the second analysis dictionary and the phonetic description of other reading candidates. Voice recognition is performed using the voice recognition dictionary in which is registered. Therefore, for example, the second
The obfuscated word "Kyoto" as an example of vocabulary registered in the analysis dictionary
Even if you accidentally say "Kyoushu", you will not be rejected and the target vocabulary "Kyoto" will be recognized as the recognition result.
Is obtained.

In one embodiment, in the voice recognition device of the second invention or the third invention, the voice recognition result has the same vocabulary and notation stored in the second analysis dictionary storage means. The reading determination means for determining whether or not the reading includes different vocabulary, and when the reading determination means determines that the vocabulary is included, the second analysis dictionary storage means for the vocabulary is read. And a reading presenting means for presenting the readings stored in.

According to this embodiment, during the voice recognition result,
When the vocabulary stored in the second analysis dictionary includes a vocabulary of a reading different from the reading of the vocabulary stored in the second analysis dictionary, the reading presenting means is used to The correct reading stored in the second analysis dictionary storage means is presented. In this way, the correct reading of the recognized vocabulary is taught to the user.

In one embodiment, in the voice recognition device of the second invention or the third invention, the reading presenting means synthesizes the reading presentation stored in the second analysis dictionary storage means. It is supposed to be done by voice.

According to this embodiment, the correct reading of the recognized vocabulary is taught to the user by the synthetic voice.

Further, the portable terminal device of the fourth invention is characterized by being equipped with the voice recognition device of the second invention or the third invention.

[0028] Normally, the mobile terminal is used when moving. In particular, when the voice recognition is performed by the mobile terminal device at a place where the user is out, if the voice is uttered with an incorrect reading and is rejected, there is no way to check the correct reading. Therefore, there may be a case where necessary information cannot be retrieved immediately.

According to the above configuration, even if the obfuscated word "Kyoto" is mistakenly uttered as "kyoshu" in the mobile terminal device, the target vocabulary "Kyoto" is obtained as a recognition result without being rejected. A recognition device is installed. Therefore, even when the user is out of the office where there is no way to check the correct reading, it becomes possible to immediately retrieve the necessary information by voice.

The portable terminal device of the fifth invention is characterized by being equipped with either one of the speech recognition dictionary creating device of the first invention and the speech recognition device of the second invention.

According to the above construction, the first portable terminal equipped with the speech recognition dictionary creating apparatus of the first invention is created in the second portable terminal equipped with the speech recognition device of the second invention. By transmitting the information of the recognized voice recognition dictionary, the voice recognition device of the second mobile terminal mistakenly changes the obfuscated word "Kyoto" as an example of the vocabulary registered in the second analysis dictionary to " Even if you say "Kyoushu", you will not be rejected, and the target vocabulary "Kyoto" will be obtained as a recognition result.

The speech recognition system of the sixth invention is
A mobile phone equipped with a server provided with the voice recognition dictionary creating apparatus of the first invention, and a voice recognition apparatus of the second invention, and having a transmitting / receiving unit for transmitting / receiving voice recognition dictionary information to / from the server. It is characterized by having a terminal.

According to the above construction, the voice recognition dictionary creating apparatus of the first invention is provided in the server. Therefore, the portable terminal can be made simpler and lighter in weight than the portable terminal equipped with the voice recognition device of the third invention. Further, by using the server as the third dictionary storage means and periodically supplementing the contents of the second analysis dictionary storage means, new words and foreign words that increase one after another, television program names that are regularly updated, etc. Thus, the user of the mobile terminal can recognize the voice without knowing the contents of the second analysis dictionary.

Further, a seventh invention is a text analysis means,
It has a reading adding means, a voice recognition dictionary creating means, and a voice recognition dictionary storing means, a text analysis step of analyzing character string information and dividing it into constituent words, and a reading addition step of giving reading to the divided constituent words. In the voice recognition dictionary creating method having a voice recognition dictionary creating step of creating a voice recognition dictionary based on the result of the text analysis and reading and storing the voice recognition dictionary in the voice recognition dictionary storage means, the text analysis by the text analysis means is: First
Of the vocabulary not stored in the first analytic dictionary storage means stored in the second analytic dictionary storage means and the first analytic dictionary stored in the analytic dictionary storage means Second, consisting of information including notation and reading
When the reading addition is performed by referring to the analysis dictionary, the second addition is made in the text analysis result in the reading addition by the reading addition unit.
When the vocabulary obtained by referring to the analysis dictionary is included, in addition to the reading obtained by referring to the second analysis dictionary, other reading candidates are added to the vocabulary. It is characterized by having done.

According to the above configuration, in the voice recognition dictionary created based on the text analysis result and the reading addition result, refer to the second analysis dictionary for the vocabulary registered in the second analysis dictionary. The dictionary information based on the phoneme notation of the reading obtained and the phoneme notation of the other reading candidates is registered. Therefore, by performing voice recognition using this voice recognition dictionary, the vocabulary “Kyobate” registered in the second analysis dictionary is mistakenly changed to “Kyoshu”.
The desired vocabulary "Kyoto" can be obtained as a recognition result without being rejected even when uttering.

A speech recognition dictionary creating program according to an eighth aspect of the present invention causes a computer to execute the text analyzing means, the reading adding means, the speech recognition dictionary creating means according to the first aspect of the invention.
It is characterized in that it functions as a voice recognition dictionary storage means, a first analysis dictionary storage means, and a second analysis dictionary storage means.

According to the above configuration, as in the case of the first aspect of the invention, the second analysis dictionary is referred to for the vocabulary registered in the second analysis dictionary in the created voice recognition dictionary. The dictionary information based on the phoneme notation of the reading obtained and the phoneme notation of other reading candidates is registered. Therefore, by performing voice recognition using this voice recognition dictionary, even if the vocabulary "Kyobate" registered in the second analysis dictionary is erroneously pronounced as "Kyoushu", it is rejected. Instead, the desired vocabulary "Kyoto" is obtained as a recognition result.

The program recording medium of the ninth invention is characterized in that the voice recognition dictionary creating program of the eighth invention is recorded.

According to the above configuration, the recorded speech recognition dictionary creating program is read out by the computer and used, so that the vocabulary registered in the second analysis dictionary is the same as in the case of the first invention. A voice recognition dictionary is created in which dictionary information based on the phoneme notation of reading and the phoneme notation of other reading candidates obtained by referring to the second analysis dictionary is registered. Therefore, by performing voice recognition using this voice recognition dictionary, even if the vocabulary "Kyobate" registered in the second analysis dictionary is erroneously pronounced as "Kyoushu", it is rejected. As a result, the desired vocabulary "Kyoto" is obtained.

[0040]

BEST MODE FOR CARRYING OUT THE INVENTION The present invention will be described in detail below with reference to the embodiments shown in the drawings.

<First Embodiment> The present embodiment relates to a voice recognition dictionary creating apparatus for creating a voice recognition dictionary that can be recognized even when a user does not know the correct reading of a recognition target vocabulary.

FIG. 1 is a block diagram showing the configuration of the speech recognition dictionary creating apparatus of this embodiment. The text analysis unit 1 analyzes the language of the input character string (text analysis) and divides it into configured morphemes. In that case, when there are a plurality of division candidates, all of them are output. It should be noted that each division candidate is given a likelihood indicating the degree of possibility of the division candidate.

The reading imparting section 2 imparts readings of the divided morphemes. When there are a plurality of readings, there are a method of outputting all of the plurality of readings and a method of focusing on one of the readings with the highest possibility.

The voice recognition dictionary creating unit 3 creates a voice recognition dictionary required for voice recognition based on the analysis result by the text analyzing unit 1 and the reading provided by the reading providing unit 2. Here, the voice recognition dictionary has a format in which the recognition vocabulary and its phoneme notation are paired and stored,
There is a format in which the occurrence chain probability of each recognition vocabulary is stored. Generally, in the case of discrete word speech recognition for uttering a word, only the former form of speech recognition dictionary is used, and in the case of continuous speech recognition for uttering a sentence, both of the former and the latter are recognized. A voice recognition dictionary is often used.

As the appearance chain probability of each recognition vocabulary used in the case of continuous speech recognition, a statistical language model typified by N-gram is used, or a binary appearance chain indicating whether or not to chain is used. The probability may be expressed and the vocabulary chain information may be described in grammar. Based on the data of the text analysis result and the assigned reading, it is possible to convert into the appearance chain probability in any of the above cases.

The voice recognition dictionary storage unit 4 stores the voice recognition dictionary created by the voice recognition dictionary creating unit 3. still,
The storage medium that constitutes the voice recognition dictionary storage unit 4 is a generally widely used storage device such as a flash memory or a hard disk. Further, the storage format in the voice recognition dictionary storage section 4 is, as described above, a format in which the recognition vocabulary and its phoneme notation are stored in pairs, and a format in which the occurrence chain probability of each vocabulary is stored. The voice recognition dictionary stored in the voice recognition dictionary storage unit 4 in this manner is referred to when performing voice recognition described later.

The first analysis dictionary storage unit 5 stores an analysis dictionary used when the text analysis unit 1 performs the text analysis. Here, the text analysis is performed using a method called morpheme analysis, but an analysis dictionary is required to perform this morpheme analysis. This analysis dictionary stores information such as notation, reading, and part-of-speech information for everyday words. Then, when performing the text analysis, the word (morpheme) of the text is identified by performing the matching process between the input text and the analysis dictionary. That is, the first analysis dictionary storage unit 5 stores a first analysis dictionary composed of information such as notation, reading, and part-of-speech information of commonly used vocabulary.

The second analysis dictionary storage unit 6 stores a second analysis dictionary composed of information such as special vocabulary notation, reading, and part-of-speech information not stored in the first analysis dictionary storage unit 5. . Except for whether the registered vocabulary is general or special, the structure in both analysis dictionary storage units 5 and 6 and the reference method from the text analysis unit are the same. Incidentally, as an example of the special vocabulary registered in the second analysis dictionary storage unit 6, there are a place name, a person name, etc. that cannot be read by a normal reading method.

The speech recognition dictionary creating apparatus having the above-described configuration operates as follows. FIG. 2 is a flowchart of a voice recognition dictionary creation processing operation performed by the voice recognition dictionary creation device shown in FIG. Hereinafter, the voice recognition dictionary creation process will be described in detail with reference to FIG. In order to perform this voice recognition dictionary creation processing, character string information for generating the voice recognition dictionary is required. Character string information is text analysis unit 1
Is input, the voice recognition dictionary creation processing operation starts.

In step S1, the text analysis unit 1 acquires a character string of one sentence from the input character string. In step S2, the text analysis unit 1 performs the text analysis. That is, the first analysis dictionary stored in the first analysis dictionary storage unit 5 and the second analysis dictionary stored in the second analysis dictionary storage unit 6 are collated by the morphological analysis processing. Then, the input character string information of one sentence is divided into word units. As described above, the first and second analysis dictionary storage units 5 and 6 store information such as word notation, reading, and part-of-speech, which are collated with the first and second analysis dictionary storage units 5 and 6. This makes it possible to know what the constituent words of the input character string are.

For example, when the character string "weather of tomorrow" is input, it is divided into morphemes of "tomorrow (noun)", "no (particle)" and "weather (noun)". Also, when the character string "Wait until come" is entered, the division result of "Kuru (verb)", "Made (particle)" and "Wait (verb)" and "Kuruma (noun)", " 2 with the division result of "(particle)" and "wait (verb)"
There are different division results, and a likelihood that represents the certainty is given to both division results.

Regarding the above morphological analysis process, the right longest matching method or the connection table is generally used.
He is familiar with literature such as "Basics of Natural Language Analysis" (Hozumi Tanaka: Sangyo Tosho 1989).

In step S3, the reading adding section 2 adds reading to each of the divided morphemes based on the text analysis result from the text analysis section 1. still,
When there are a plurality of readings, it is possible to output all the readings, or it is possible to output some readings from the most probable one according to the likelihood of reading. In the case of the above-mentioned example, the divided word “tomorrow” has two types of readings, “tomorrow” and “tomorrow”, and the likelihood is given to each of them.

In step S4, the reading adding section 2 determines whether or not the morpheme input from the text analysis section 1 includes a vocabulary registered in the second analysis dictionary. This determination is made by, for example, setting a flag on the text analysis result based on the second analysis dictionary. As a result, if it is included, the process proceeds to step S5, and if it is not included, the process proceeds to step S6. In step S5, the reading adding unit 2 causes the second
Regarding the vocabulary included in the analysis dictionary, the correspondence between the divided words by the second analysis dictionary and the reading and the correspondence between the divided word candidates including the analysis result by the first analysis dictionary and the reading candidates correspond to each other. Recorded on the table. A specific example will be described below.

For example, the place name "Kyoto" mentioned above is read as "Kyobate". However, since it is not a general word, it is often not registered in an ordinary analysis dictionary. That is, in the case of the present embodiment, the word “Kyoto” is not registered in the first analysis dictionary storage unit 5, but is registered in the second analysis dictionary storage unit 6. On the other hand, it is assumed that the vocabulary “K” and the vocabulary “End” are registered in the first analysis dictionary storage unit 5.

In this case, when the character string “Kyoto” is input to the text analysis unit 1, the text analysis unit 1 matches the “Kyoto” registered in the second analysis dictionary storage unit 6. Then, the word "Kyoto (noun)" is obtained. Then, the reading adding unit 2 adds the reading “Kyoubate”. Here, since the word “Kyobate: noun” is a vocabulary registered in the second analysis dictionary, the analysis result using the second analysis dictionary is “Kyobate:
Noun "and the analysis result candidates using the first analysis dictionary," Kyo: noun "/" End: noun "or" Kyo: noun "/" End: " The "noun" is recorded in the correspondence table as shown in FIG.

If the analysis dictionary storage unit is not divided into the first analysis dictionary storage unit 5 and the second analysis dictionary storage unit 6, the expression "Kyoto" is called "Kyobate". Only the reading can be obtained, "Kyoshu" and "Kyoto End"
Is never obtained. As in the present embodiment, the reading imparting unit 2 generates a reading based on the reading obtained by referring to the first analysis dictionary storage unit 5 and the reading obtained by referring to the second analysis dictionary storage unit 6. By doing so, you can get the three readings of "Kyobate", "Kyoshu", and "Kyoendo".

In step S6, the voice recognition dictionary creating section 3 generates a voice recognition dictionary based on the information of the text analysis result and the reading addition result including the contents of the correspondence table. Then, the generated voice recognition dictionary is stored in the voice recognition dictionary storage unit 4. Step S7
Then, the text analysis unit 1 determines whether or not there is a next sentence in the input character string. As a result, if there is, the process returns to the step S1 to proceed to the acquisition of the character string of the next sentence. If not, the voice recognition dictionary creation processing operation is terminated.

As described above, in the present embodiment, the first analysis dictionary storage unit 5 for storing the first analysis dictionary in which information such as notation, reading, and part-of-speech information of vocabulary used daily is stored.
And a second analysis dictionary storage unit 6 that stores a second analysis dictionary in which information such as special vocabulary notation, reading, and part-of-speech information is registered. Then, the text analysis unit 1 performs the morpheme analysis using both analysis dictionary storage units 5 and 6, and the reading addition unit 2 adds the reading to the morpheme based on the morpheme analysis result. The voice recognition dictionary creating unit 3 creates a voice recognition dictionary based on the analysis result and the reading.

At this time, when the input morpheme contains the registered vocabulary of the second analysis dictionary, the reading imparting unit 2 makes a division word "Kyoto" for the vocabulary by the analysis using the second analysis dictionary. In addition to the pair "" and its reading "Kyobate", other divided word candidates "Kyo / Shu" and its reading candidate "Kyo / Shu", including the analysis results using the above first analysis dictionary,
Record the pair “Today / End” in the correspondence table. Then, the voice recognition dictionary creating section 3 creates a voice recognition dictionary based on the recorded contents of the correspondence table.

As a result, the speech recognition dictionary has, for example, the recognition vocabulary “Kyoto” and the phoneme notation “Kyobate”, “Kyoshu”,
It will be created based on the correspondence with the “Kyoto”. Therefore, by performing voice recognition using the voice recognition dictionary created by the voice recognition dictionary creating unit 3, even when the notation "Kyoto" is pronounced as "Kyo-end", it is correctly recognized as "Kyoto". You can do it.

<Second Embodiment> The present embodiment relates to the speech recognition dictionary creating apparatus according to the first embodiment, which is provided with the location acquisition unit for automatically acquiring the second analysis dictionary. .

FIG. 4 is a block diagram showing the structure of the speech recognition dictionary creating apparatus of this embodiment. The text analysis unit 11, the reading addition unit 12, the voice recognition dictionary creation unit 13, the voice recognition dictionary storage unit 14, the first analysis dictionary storage unit 15, and the second analysis dictionary storage unit 16 are the same as those in the first embodiment shown in FIG. It is the same as the text analysis unit 1, reading addition unit 2, voice recognition dictionary creation unit 3, voice recognition dictionary storage unit 4, first analysis dictionary storage unit 5 and second analysis dictionary storage unit 6 shown in FIG. Omit it.

While the general vocabulary is registered in the first analysis dictionary storage unit 15, the special vocabulary is registered in the second analysis dictionary storage unit 16 according to the first embodiment. It is similar to the case. Here, the special vocabulary includes, for example, highly specialized vocabulary, infrequently appearing unfamiliar vocabulary, abbreviations, new words, and difficult-to-read words.

By the way, with respect to the highly specialized vocabulary, abbreviations, new words, etc., new vocabulary constantly appears with the passage of time. Therefore, when this newly appearing special vocabulary is not constantly registered in the second analysis dictionary storage unit 16,
Even if the newly appearing special vocabulary is input to the text analysis unit 11, there is a high possibility that the reading is not correctly given. There is also a technique of estimating the reading when the reading cannot be accurately given and it is determined as an unknown word. However, the accuracy of reading estimation by this reading estimation technique is not so high. Therefore, in order to accurately add reading, it is necessary to constantly register the newly appearing special vocabulary in the second analysis dictionary storage unit 16.

Therefore, in this embodiment, a dictionary acquisition unit 17 is provided to store information such as special vocabulary notation, reading, and part-of-speech information stored in the second analysis dictionary storage unit 16 in the third dictionary storage unit. It is acquired by the dictionary acquisition unit 17 from a means (not shown). By doing so, if a new word appears, such as a new word, the second analysis dictionary storage unit 16
You can additionally register with.

By doing so, it becomes possible to register words that are updated daily, such as electronic program information such as TV program names and music title names, in the second analysis dictionary storage section 16. Therefore, if the dictionary acquisition unit 17 regularly registers a newly appearing special vocabulary in the second analysis dictionary storage unit 16, the recognition target vocabulary registered in the second analysis dictionary storage unit 16 is used. Even if the person does not know, a voice recognition dictionary capable of recognizing the input voice can be created and stored in the voice recognition dictionary storage unit 14.

Here, the method of acquiring the special vocabulary by the third dictionary storage means and the dictionary acquisition unit 17 is not particularly limited. For example, a floppy (registered trademark) disk or CD (compact disk) -R
There are a method of acquiring from a medium such as OM (Read Only Memory), a method of downloading from a network, and a method of using a mechanism such as teletext. In any case, if a third vocabulary storage means in which a new vocabulary is registered by a vocabulary information provider is prepared, the new vocabulary can be acquired from the third dictionary storage means by the dictionary acquisition unit 17 and used. You can do it.

As described above, in the present embodiment,
The dictionary acquisition unit 17 is provided, and the dictionary acquisition unit 17 acquires the special vocabulary stored in the second analysis dictionary storage unit 16 from the third dictionary storage unit. Therefore, if the dictionary acquisition unit 17 periodically acquires new vocabulary information from the third dictionary storage unit and registers it in the second analysis dictionary storage unit 16, a TV program name, a music title name, etc. It is possible to create a voice recognition dictionary that can recognize even a word that the user does not know is registered in the second analysis dictionary storage unit 16 because the word is updated every day.

<Third Embodiment> The present embodiment relates to a voice recognition device equipped with the voice recognition dictionary creating device according to the first embodiment.

FIG. 5 is a block diagram showing the configuration of the speech recognition apparatus of this embodiment. Text analysis unit 2
1, the reading adding unit 22, the voice recognition dictionary creating unit 23, the first analysis dictionary storage unit 24, and the second analysis dictionary storage unit 25 are the text analysis unit 1 shown in FIG. 1 in the first embodiment.
It is the same as the reading providing unit 2, the voice recognition dictionary creating unit 3, the first analysis dictionary storage unit 5, and the second analysis dictionary storage unit 6, and constitutes a voice recognition dictionary creating device 26. Then, the voice recognition dictionary created by the voice recognition dictionary creating device 26 is stored in the voice recognition dictionary storage unit 27. The detailed description of the voice recognition dictionary creating device 26 and the voice recognition dictionary storage unit 27 will be omitted.

The speech recognition section 31 is composed of an acoustic analysis section 28, a likelihood calculation section 29 and a matching processing section 30 and matches the input speech with the words registered in the speech recognition dictionary storage section 27. Recognize and output the recognition result. The outline will be described below.

The acoustic analysis unit 28 includes a microphone (not shown).
The voice input from is converted into a digital waveform, frequency-analyzed for each short time interval (frame), and converted into a vector series of parameters representing a spectrum. An expression method such as LPC (linear prediction analysis) mel cepstrum is used for frequency analysis. The likelihood calculating unit 29 uses the acoustic model (HMM) for the obtained input voice parameter vector.
(Hidden Markov model) etc. is applied to calculate the likelihood for each phoneme. The matching processing unit 30 matches the phonological likelihood (similarity) series with all the items (words) stored in the speech recognition dictionary storage unit 27, and calculates the score of each word. Then, the word with a high score is output as the recognition result. The speech recognition method is detailed in the literature such as "Digital Speech Processing" (Furui: Tokai University Press, 1985).

In this case, as described in the first embodiment, the voice recognition dictionary of the voice recognition dictionary storage unit 27, for example, for the place name "Kyoto" which is a difficult-to-read word, has its correct phoneme notation " In addition to "Kyobate", the wrong phoneme notations "Kyoshu" and "Kyo-end" are also associated with each other. Therefore, when the speaker utters "Kyobate" into the microphone, the recognition result "Kyoto"
Can be obtained. In addition to that, even if the user accidentally utters "Kyoshu" or "Kyoto", the correct recognition result "Kyoto" can be obtained.

That is, according to the present embodiment, the input voice is used as the target vocabulary even if the reading of a place name or a person's name, which is a difficult-to-read word, is erroneously memorized or the accurate reading is not understood. Can be recognized. Therefore, the present embodiment is
When applied to a map search display device or the like that displays a searched map by inputting a place name with many obfuscated words by utterance, it can be used very effectively.

In this embodiment, the voice recognition dictionary creation device 26 is mounted on the voice recognition device. However, the present invention is not limited to this, and a voice recognition device is provided independently of the voice recognition dictionary creation device 26, and a voice created by the voice recognition dictionary creation device according to the first and second embodiments described above. It does not matter if the recognition dictionary is stored in the voice recognition dictionary storage unit 27.

<Fourth Embodiment> In the present embodiment, the voice recognition dictionary creating apparatus according to the first embodiment is installed, and when a difficult-to-read word is uttered by mistake, correct reading is presented and taught. Voice recognition device.

FIG. 6 is a block diagram showing the configuration of the speech recognition apparatus of this embodiment. Text analysis unit 4
1, the reading imparting unit 42, the voice recognition dictionary creating unit 43, the first analysis dictionary storage unit 44, and the second analysis dictionary storage unit 45 are the text analysis unit 1 shown in FIG. 1 in the first embodiment.
It is the same as the reading adding unit 2, the voice recognition dictionary creating unit 3, the first analysis dictionary storage unit 5, and the second analysis dictionary storage unit 6, and constitutes a voice recognition dictionary creating device 46. Then, the voice recognition dictionary created by the voice recognition dictionary creating device 46 is stored in the voice recognition dictionary storage unit 47. The detailed description of the voice recognition dictionary creating device 46 and the voice recognition dictionary storage unit 47 will be omitted.

The voice recognition section 48 has the same structure as the voice recognition section 31 shown in FIG. 5 in the third embodiment. Then, the input speech is subjected to acoustic analysis to be converted into a vector series of parameters, and an acoustic model is applied to the parameter vector to perform likelihood calculation for each phoneme, and a phonological likelihood series and a speech recognition dictionary storage unit 47. The word of each word is calculated by collating with the total number of words, and the word with the highest score is output as the recognition result.

The reading determination unit 49 receives the result of the voice recognition from the voice recognition unit 48, and the reading is the same as the vocabulary stored in the second analysis dictionary storage unit 45, but the reading is not performed. Determine if different vocabularies are included. The reading presentation unit 50 receives the determination result of the reading determination unit 49, and when the determination result is “true”, the second analysis dictionary storage unit 45.
The reading of the vocabulary stored in is presented. That is, when the wrong reading of the vocabulary stored in the second analysis dictionary storage unit 45 is input (spoken) to the voice recognition unit 48, the correct reading of the vocabulary is presented and taught to the user.

The voice recognition device having the above configuration operates as follows. FIG. 7 is a flowchart of the voice recognition processing operation executed by the voice recognition unit 48, the reading determination unit 49, and the reading presentation unit 50. The voice recognition processing operation will be described below with reference to FIG. Microphone
When a voice is input to the voice recognition unit 48 (not shown), the voice recognition processing operation starts.

In step S11, the voice recognition unit 48 converts the input voice into a digital waveform, frequency-analyzes each frame, and converts it into a vector series of parameters representing a spectrum (acoustic analysis). Further, the acoustic model is applied to the parameter vector to perform likelihood calculation for each phoneme (likelihood calculation). Then, the phonological likelihood series is compared with all the words registered in the voice recognition dictionary storage unit 47 to calculate the score of each word.
(Matching process).

In step S12, based on the result of the voice recognition from the voice recognition unit 48 by the reading determination unit 49, the vocabulary and the notation stored in the second analysis dictionary storage unit 45 are the same. It is determined whether or not a vocabulary, which has a different reading, is included. That is, the voice recognition result is as shown in FIG.
It is determined whether or not it is included in the “other candidates” of the correspondence table shown in FIG. As a result, if included, the process proceeds to step S13, and if not, the process proceeds to step S14.

Here, the fact that the voice recognition result is included in the "other candidates" of the correspondence table means, for example,
A person who saw the notation "Kyoto" uttered "Kyoshu",
It corresponds to the case where it is directly recognized as "Kyushu". Then, the determination in this step is, for example,
When the voice recognition dictionary creating unit 43 creates a voice recognition dictionary based on the above correspondence table, of the phoneme notations “Kyobate”, “Kyoshu”, and “Kyo End” associated with the recognition vocabulary “Kyoto” Phoneme notation "Kyoshu", "Kyo End" included in "Other candidates" in the above correspondence table
It can be realized by setting a flag. Alternatively, it can be realized by setting a flag to the appearance chain probability based on the phoneme notation included in “other candidates” in the correspondence table among the appearance chain probabilities of the recognition vocabulary.

That is, when the speech recognition unit 48 performs the collation processing, it is calculated when the speech recognition dictionary 48 collates with the phoneme notation or the word having the appearance chain probability flagged as described above. It is only necessary to add information indicating that to the score.

In step S13, the phonetic notation "Kyoubate" of the second analysis dictionary vocabulary corresponding to the phoneme notation "Kyoshu" included in the "other candidates" of the correspondence table is obtained by the reading presentation unit 50. And is returned to the voice recognition unit 48. In this case, the phoneme notation “Kyobate” in the second analytic dictionary vocabulary refers specifically to the inside of the voice recognition dictionary storage storage unit 47 and is associated with the recognition result “Kyoto”. It is performed by obtaining a phoneme notation "Kyobate" in which the above flag is not set among the notations.

In step S14, the speech recognition unit 48 outputs the word with the high score calculated in step S11 as the recognition result. At this time, when the phonetic notation of the second analysis dictionary vocabulary of the correspondence table is returned from the reading presentation unit 50, the phonemic notation is also output and displayed. In this way, by outputting the recognition result “Kyoto” together with the original reading “Kyobate”, the user can read the “Kyoshu” reading of the vocabulary “Kyoto” input by voice.
Is a mistake and can teach that the true reading is "Kyoubate".

Here, the utterance "Kyoshu" is input to the voice recognition unit 48, and the recognition results "Kyoto", "nostalgia",
When there are a plurality of candidates such as “training”, the voice recognition unit 48 once displays a plurality of candidates “Kyoto”, “nostalgia”, and “training”,
Let the user select one of the candidates. As a result, when the recognition candidate "Kyoto" is selected, the processing by the reading determination unit 49 and the reading presentation unit 50 as described above may be performed.

As described above, in the present embodiment, in addition to the voice recognition section 48, the reading determination section 49 and the reading presentation section 50 are provided. Then, the reading determination unit 49
In the voice recognition, the second analysis dictionary storage unit 4
It is determined whether or not a vocabulary stored in No. 5 has the same notation as the vocabulary but a different reading. When it is determined that the vocabulary is included, the reading presentation unit 5
The phoneme notation of the second analysis dictionary of the vocabulary whose reading is determined to be different depending on 0 is presented to the voice recognition unit 48 together with the recognition result.

Therefore, the second analysis dictionary storage section 45 is used.
When the vocabulary registered in is uttered with a reading different from the correct reading and the utterance is correctly recognized by the voice recognition unit 48, the correct reading can be output and displayed together with the recognition result, and the user can be taught. You can do it.

In the present embodiment, the reading presenting section 50 obtains the phoneme notation “Kyobate” of the vocabulary of the second analysis dictionary and returns it to the voice recognizing section 48.
The output and display together with the result of voice recognition are shown by 8. However, the present invention is not limited to this. For example, the reading and presenting unit 50 may be provided with a voice synthesizing unit to output the voice recognition result by the voice recognizing unit 48 in synchronization with the output display of the voice recognition result.

Further, in the present embodiment, the voice recognition device is equipped with the voice recognition dictionary creating device 46. However, the present invention is not limited to this,
A voice recognition device is provided independently of the voice recognition dictionary creation device 46, and the voice recognition dictionary created by the voice recognition dictionary creation device in the first and second embodiments is stored in the voice recognition dictionary storage unit 47. It doesn't matter.

Further, in the present embodiment, the judgment by the reading judging unit 49 and the acquisition of the correct reading by the reading presenting unit 50 are included in the “other candidates” of the correspondence table of each recognized vocabulary in the voice recognition dictionary. This is done by setting a flag in the phoneme notation described and referring to this flag. However, it does not matter if the correspondence table is directly referred to. However, in that case, since the correspondence table needs to be shared by the voice recognition dictionary creation device 46, the reading determination unit 49, and the reading presentation unit 50, the voice recognition device is equipped with the voice recognition dictionary creation device 46. Need to be

In the third and fourth embodiments, the case where the voice recognition dictionary creating apparatus according to the first embodiment is installed is described as an example, but the second embodiment is described. It does not matter even if the voice recognition dictionary creating device in the above form is installed.

The voice recognition devices according to the third and fourth embodiments described above can exert their effects by being mounted on a portable terminal device. Usually, the mobile terminal is used when moving. And, especially when going out and searching a map by voice input with the above-mentioned mobile terminal device and displaying it, for example, a person who thinks that the place name "Kyobate" is "kyoshu" is You will be uttered with the wrong reading. In the case of the present mobile terminal device, even in that case, it is not rejected and the map of the target place name "Kyoto" is displayed.

On the other hand, in the case of a mobile terminal device equipped with a map search device using a conventional voice recognition device, for example, if the place name "Kyoto" is uttered as "Kyoushu", it is rejected. However, there is no way to check the correct reading while on the go, which makes it impossible to display the "Kyoto" map.

In addition, the above-mentioned portable terminal device is replaced with the above-mentioned first and second
The first mobile terminal device equipped with the voice recognition dictionary creation device according to the embodiment, and the voice recognition dictionary storage unit, the voice recognition unit, the reading determination unit, and the reading presentation unit according to the third and fourth embodiments are mounted. It is also possible to configure it with the second mobile terminal device, and to provide a transceiver for transmitting and receiving information including voice recognition dictionary information between both mobile terminal devices to both mobile terminal devices. By doing so, the voice recognition dictionary information created by the voice recognition dictionary creating device of the first mobile terminal is used as the second voice recognition dictionary information.
It can be transmitted to the mobile terminal and stored in the voice recognition dictionary storage unit of the second mobile terminal.

The voice recognition dictionary creating device for creating the voice recognition dictionary of the voice recognition device is the voice recognition dictionary creating device in the second embodiment, and the voice recognition dictionary creating device is provided in the server. In addition, the mobile terminal has a second
It is also possible to provide a voice recognition dictionary storage unit, a voice recognition unit (reading determination unit, reading presentation unit), and a transceiver for transmitting and receiving voice recognition dictionary information to and from the server. As described above, by configuring the voice recognition system with the server and the mobile terminal device, the mobile terminal device can have a simple structure and can be reduced in weight. Further, by using the server as the third dictionary storage means, the contents of the second analysis dictionary storage section in the server are supplemented periodically, and new words and foreign words that are increasing one after another and regularly updated. A voice recognition dictionary that can handle TV program names etc.
It can be acquired by the transceiver.

By the way, the text analyzing units 1, 11, 21, 41 and the reading adding units 2, 12, 2 in each of the above embodiments.
2, 42, voice recognition dictionary creation unit 3, 13, 23, 43, voice recognition dictionary storage unit 4, 14, 27, 47, first analysis dictionary storage unit 5, 15, 24, 44, second analysis dictionary storage unit 6,16,2
The functions of 5, 45 are realized by the voice recognition dictionary creating program recorded in the program recording medium.
The program recording medium in each of the above embodiments is
The program medium is a ROM (not shown). Alternatively, it may be a program medium loaded in an external auxiliary storage device and read. In any case, the program reading means for reading the voice recognition dictionary creating program from the program medium may have a configuration of directly accessing and reading the program medium, or a RAM (random access memory). You may have the structure which downloads to the program storage area provided in (not shown), accesses the said program storage area, and reads it. The download program for downloading from the program medium to the program storage area of the RAM is assumed to be stored in the main body device in advance.

Here, the program medium is configured to be separable from the main body side, and is a tape system such as a magnetic tape or a cassette tape, a magnetic disk such as a floppy disk or a hard disk, a CD-ROM, an MO (magneto-optical) disk. , MD (mini disk), DVD (digital versatile disk) and other optical disk systems, IC (integrated circuit) cards, optical cards and other card systems, mask ROM, EPRO
M (UV erasable ROM), EEPROM (electrically erasable R)
OM), including flash ROM and other semiconductor memory systems,
It is a medium that carries the program fixedly.

Further, when the voice recognition dictionary creating apparatus in each of the above embodiments has a configuration that includes a modem and is connectable to a communication network including the Internet, the program medium is downloaded from the communication network. It may be a medium that supports the program in a fluid manner. In this case, the download program for downloading from the communication network is stored in the main body device in advance. Alternatively, it is assumed that the program is installed from another recording medium.

It should be noted that what is recorded on the recording medium is not limited to the program, and data can be recorded.

[0103]

As is apparent from the above, the voice recognition dictionary creating apparatus of the first invention stores the first analysis dictionary composed of information including vocabulary notation and reading as a text analysis dictionary. The reading includes a first analysis dictionary storage unit and a second analysis dictionary storage unit in which a second analysis dictionary composed of information including notations and readings of vocabulary not stored in the first analysis dictionary storage unit is stored. When the adding means includes the vocabulary obtained by referring to the second analysis dictionary in the text analysis result, in addition to the reading obtained by referring to the second analysis dictionary with respect to the vocabulary. Since other reading candidates are also added, the phoneme of the reading obtained by referring to the second analysis dictionary with respect to the vocabulary is included in the speech recognition dictionary created based on the analysis result and the reading addition result. Notation and Dictionary information based on the phoneme notation of other reading candidate is registered. Therefore, by performing voice recognition using this voice recognition dictionary, for example, even if the obfuscated word “Kyobate” registered in the second analysis dictionary is erroneously pronounced as “Kyoshu”, it is rejected. Therefore, the desired vocabulary "Kyoto" can be obtained as a recognition result.

That is, according to the present invention, even when the user does not know the correct reading of the recognition target word,
It is possible to create a voice recognition dictionary that can recognize the input voice.

Further, in the voice recognition dictionary creating apparatus according to the first embodiment, the dictionary acquisition means acquires the contents of the second analysis dictionary stored in the second analysis dictionary storage means from the third dictionary storage means. By providing the third dictionary storage means in which a new vocabulary is registered by a vocabulary information provider, a newly appearing vocabulary can always be additionally registered in the second analysis dictionary storage means. By doing so, even if the user does not know the recognition target vocabulary registered in the second analysis dictionary, it is possible to create a voice recognition dictionary that can recognize the input voice.

That is, according to the present invention, even if the user does not know the registered recognition target word,
It is possible to create a voice recognition dictionary that can recognize the input voice.

Further, since the voice recognition device of the second invention recognizes the input voice by collating with the registered vocabulary of the voice recognition dictionary created by the voice recognition dictionary creation device of the first invention, for example, Even if the obfuscated word "Kyobate" registered in the second analysis dictionary of the voice recognition dictionary creation device is mistakenly uttered as "Kyoshu", it is not rejected and the target is obtained as a recognition result. You can get the vocabulary “Kyoto”.

The voice recognition device of the third invention is equipped with the voice recognition dictionary creation device of the first invention described above, and collates with the registered vocabulary of the voice recognition dictionary created by this voice recognition dictionary creation device. Since the input voice is recognized by performing the input voice, it is possible to create a voice recognition dictionary capable of recognizing the input voice even when the user does not know the correct reading of the recognition target word.

The voice recognition apparatus of the first embodiment reads whether or not a vocabulary stored in the second analysis dictionary storage means has the same notation but is read differently from the voice recognition result. When it is determined by the determination means and the vocabulary is included, the reading presenting means presents the reading stored in the second analysis dictionary storage means with respect to the vocabulary. For users who mistakenly uttered the registered obfuscated word "Kyobate" as "Kyoshu" and obtained the correct recognition result "Kyoto",
You can teach by presenting the correct reading "Kyobate".

Further, in the voice recognition apparatus according to the first embodiment, the reading presentation means is adapted to present the reading stored in the second analysis dictionary storage means by the synthetic voice, so that the user can be read. The correct reading of the recognition vocabulary can be presented by synthetic speech, and the display contents of the speech recognition result can be simplified.

Further, since the portable terminal device of the fourth invention is equipped with the voice recognition device of the second invention or the third invention, it can be used even when out of the office where there is no way to check the correct reading.
The voice makes it possible to search for necessary information immediately and easily.

Further, since the portable terminal device of the fifth invention is equipped with either one of the voice recognition dictionary creating apparatus of the first invention and the voice recognition apparatus of the second invention, the voice recognition dictionary creating The information of the created voice recognition dictionary can be transmitted from the first mobile terminal device equipped with the device to the second mobile terminal device equipped with the voice recognition device. Therefore, the second mobile terminal can be made simpler and lighter in weight than the mobile terminal equipped with the voice recognition device of the third invention.

The speech recognition system of the sixth invention is
A mobile terminal equipped with a server provided with the voice recognition dictionary creation device of the first invention, and a voice recognition device of the second invention and having a transmitting / receiving means for transmitting / receiving voice recognition dictionary information to / from the server. Since the portable terminal device is provided with a device, the portable terminal device can be made simpler in structure and lighter in weight than the portable terminal device equipped with the voice recognition device of the third invention.

Further, by using the server as the third dictionary storage means and periodically replenishing the contents of the second analysis dictionary storage means from the server, new words and foreign words that are increasing one after another and regularly updated. It becomes possible for the user of the mobile terminal device to recognize the displayed TV program name and the like by voice without knowing the contents of the second analysis dictionary.

Further, the speech recognition dictionary creating method according to the seventh aspect of the present invention includes a first analysis dictionary composed of information including notations and readings of vocabulary stored in the first analysis dictionary storage means and a second analysis dictionary storage means. A second stored information that includes notation and reading of a vocabulary not stored in the first analysis dictionary storage means.
When the text analysis is performed by referring to the analysis dictionary and the reading is added to the divided constituent words, when the vocabulary registered in the second analysis dictionary is included in the text analysis result, Adds other reading candidates in addition to the reading obtained by referring to the second analysis dictionary for the vocabulary. Therefore, the created speech recognition dictionary includes the second analysis for the vocabulary. The dictionary information based on the phoneme notation of reading and the phoneme notation of other reading candidates obtained by referring to the dictionary is registered. Therefore, by performing voice recognition using this voice recognition dictionary, for example, even if the obfuscated word “Kyobate” registered in the second analysis dictionary is erroneously pronounced as “Kyoshu”, it is rejected. Therefore, the desired vocabulary "Kyoto" can be obtained as a recognition result.

That is, according to the present invention, even if the user does not know the correct reading of the recognition target word,
A voice recognition dictionary capable of recognizing input voice can be created.

The voice recognition dictionary creating program according to the eighth aspect of the present invention causes a computer to execute the text analyzing means, the reading adding means, the voice recognition dictionary creating means according to the first aspect of the invention.
Voice recognition dictionary storage means, first analysis dictionary storage means and second
Since it functions as the analysis dictionary storage means, even if the vocabulary "Kyobate" registered in the second analysis dictionary is mistakenly pronounced as "Kyoushu" as in the case of the first invention. It is possible to create a voice recognition dictionary that can obtain the target vocabulary “Kyoto” as a recognition result without being rejected.

Further, since the program recording medium of the ninth invention records the voice recognition dictionary creating program of the eighth invention, the voice recognition dictionary creating program can be read out by a computer and used. 1
As in the case of the invention of No. 1, even if the vocabulary “Kyobate” registered in the second analysis dictionary is erroneously pronounced as “Kyoshu”, it is not rejected and the result of the recognition is It is possible to create a voice recognition dictionary that can obtain the vocabulary “Kyoto”.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a voice recognition dictionary creating device of the present invention.

FIG. 2 is a flowchart of a voice recognition dictionary creation processing operation performed by the voice recognition dictionary creation device shown in FIG.

FIG. 3 is a diagram showing an example of contents of a correspondence table recorded by a reading imparting unit in FIG.

FIG. 4 is a block diagram showing a configuration of a voice recognition dictionary creating apparatus different from that of FIG.

FIG. 5 is a block diagram showing a configuration of a voice recognition device of the present invention.

FIG. 6 is a block diagram showing a configuration of a voice recognition dictionary creation device different from that of FIG.

7 is a flowchart of a voice recognition processing operation executed by a voice recognition unit, a reading determination unit, and a reading presentation unit in FIG.

[Explanation of symbols]

1, 11, 21, 41 ... Text analysis part, 2,12,22,42 ... Reading section, 3, 13, 23, 43 ... Voice recognition dictionary creation unit, 4, 14, 27, 47 ... Voice recognition dictionary storage unit, 5, 15, 24, 44 ... First analysis dictionary storage unit, 6, 16, 25, 45 ... Second analysis dictionary storage unit, 17 ... Dictionary acquisition unit, 26,46 ... Voice recognition dictionary creation device, 28 ... Acoustic analysis unit, 29 ... Likelihood calculator, 30 ... Collation processing unit, 31, 48 ... Voice recognition unit, 49 ... a reading determination unit, 50 ... Reading presentation section.

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁷ Identification code FI theme code (reference) G10L 3/00 561D R

Claims (12)

[Claims] 1. An input text is analyzed by a text analysis means, a reading is added to the analyzed constituent words by a reading adding means, and a voice recognition is made by a voice recognition dictionary creating means based on the analysis result and the reading addition result. In a voice recognition dictionary creating device that creates a dictionary and stores the created voice recognition dictionary in a voice recognition dictionary storage means, the dictionary is referred to when the text is analyzed by the text analysis means, and the vocabulary is written and read. A first analysis dictionary storage unit that stores a first analysis dictionary that includes information including: and a dictionary that is referred to when the text is analyzed by the text analysis unit, and is not stored in the first analysis dictionary storage unit. Second, consisting of information including vocabulary and reading
The reading analysis unit includes a second analysis dictionary storage unit that stores an analysis dictionary, and the reading adding unit includes a vocabulary obtained by referring to the second analysis dictionary in the text analysis result of the text analysis unit. In the case where the vocabulary is present, the voice recognition dictionary creating apparatus is characterized in that, in addition to the reading obtained by referring to the second analysis dictionary, other reading candidates are added. 2. The voice recognition dictionary creation device according to claim 1, wherein the dictionary acquisition means for acquiring the contents of the second analysis dictionary stored in the second analysis dictionary storage means from the third dictionary storage means. An apparatus for creating a voice recognition dictionary, comprising: 3. A voice recognition device for recognizing an input voice by matching with a vocabulary registered in a voice recognition dictionary by a matching means, wherein the voice recognition dictionary is one of claims 1 and 2. A voice recognition device, which is a voice recognition dictionary created by the voice recognition dictionary creation device described in 1. 4. A voice recognition dictionary creating apparatus according to claim 1 or 2 is mounted, and an input voice is used as a vocabulary registered in a voice recognition dictionary storing means in the voice recognition dictionary creating apparatus. A voice recognition device characterized by performing verification by means of verification means. 5. The voice recognition device according to claim 3 or 4, wherein the voice recognition result includes a vocabulary having the same notation as the vocabulary stored in the second analysis dictionary storage means but different reading. When it is determined by the reading determination unit that determines whether or not the vocabulary is included, and the above-mentioned vocabulary is included, the vocabulary is stored in the second analysis dictionary storage unit. A voice recognition device comprising a reading presenting means for presenting a reading. 6. The voice recognition apparatus according to claim 5, wherein the reading presenting means presents the reading stored in the second analysis dictionary storage means with synthetic speech. Characteristic voice recognition device. 7. A mobile terminal equipped with the voice recognition device according to claim 3. 8. A voice recognition dictionary creating apparatus according to claim 1 or 2 and a voice recognition apparatus according to any one of claims 3, 5, and 6 are mounted. A mobile terminal device characterized by the above. 9. A server provided with the voice recognition dictionary creating device according to claim 1 or 2, and the voice recognition device according to any one of claims 3, 5, and 6. A voice recognition system comprising: a mobile terminal device which is mounted and has a transmitting / receiving unit for transmitting / receiving voice recognition dictionary information to / from the server. 10. A text analysis step for analyzing character string information and dividing it into constituent words, and a text analyzing means, a reading adding means, a speech recognition dictionary creating means and a speech recognition dictionary storing means, and a divided constituent word. In a voice recognition dictionary creating method, which has a reading adding step of adding reading, and a voice recognition dictionary creating step of creating a voice recognition dictionary based on a result of the text analysis and the reading adding and storing the voice recognition dictionary in the voice recognition dictionary storage means, In the text analysis by the text analysis means, the first analysis dictionary composed of information including notation and reading of vocabulary stored in the first analysis dictionary storage means, and the first analysis stored in the second analysis dictionary storage means. The reading is performed by referring to a second analysis dictionary that includes information including vocabulary notations and readings that are not stored in the dictionary storage means. In the reading addition by the giving means, when the vocabulary obtained by referring to the second analysis dictionary is included in the text analysis result, the second analysis dictionary is referred to for the vocabulary. A method for creating a voice recognition dictionary, characterized in that other reading candidates are added in addition to the obtained reading. 11. A computer is caused to function as the text analysis means, reading imparting means, voice recognition dictionary creating means, voice recognition dictionary storage means, first analysis dictionary storage means and second analysis dictionary storage means according to claim 1. A featured voice recognition dictionary creation program. 12. A computer-readable program recording medium on which the voice recognition dictionary creating program according to claim 11 is recorded.