JPH09325789A - Voice synthesizer and information retrieving system provided with synthesizer - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce the occurrence of errors in an information transmission by selecting the reading, through which the inscription of a word is understandable, from an inscription pronunciation information if a same pronunciation different inscription word exists and conducting a large vocabulary reading. SOLUTION: A pronunciation information dictionary 15 stores the pronunciation information indicating how to read words. A voice synthesis section 14 receives character code columns, extracts a word from the columns and retrieves the pronunciation information of the dictionary from the inscription of the extracted word. In this pronunciation information, a flag is provided for every word to indicate the presence or the absence of a same pronunciation different inscription word. If the flag exists, the section 14 conducts a large vocabulary reading. Moreover, in addition to the pronunciation, the dictionary 15 stores the inscription pronunciation information to indicate the reading through which the inscription of the word is understandable. If the flag is associated with a word, the section 14 refers to the inscription pronunciation information of the dictionary 15 for every inscribed character, adds the reading, through which the inscription of the word is understandable, with a proper pause after the reading of the word selected from the pronunciation information and conducts a large vocabulary reading.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a voice synthesizing device and an information retrieval system including the voice synthesizing device, which can be applied to all scenes where a human needs to obtain some voice information. The present invention relates to a voice synthesizer effective for reading a place name in the system and a person / company name search service via a telephone, and an information search system including the voice synthesizer.

[0002]

2. Description of the Related Art There is a review technique as a technique for reading some information including notation so that it can be recognized by humans. In this editing technique, the rule-based speech synthesizer generally has a dictionary in which reading information is written for each word, and the words are read out using this dictionary. In this rule-based speech synthesizer, if the number of words in the dictionary is increased, the number of words that can be correctly read can be increased accordingly.

[0003]

However, in consideration of the person who uses the speech synthesis output, it is not always possible to receive the necessary information from the speech output that is read out correctly. For example, if you pronounce the place name "Kutchan" as "Kutchan", you do not know the kanji of "Kutchan", so check the location of the place name "Kutchan" later when you look at the map or see the road guide. Can be extremely difficult. In addition, it is considered that everyone has the experience of inquiring whether it is "Ito" or "Ito" when "Ito" is pronounced in a telephone agency.

Therefore, the present invention has been made in view of the above, and an object thereof is to pronounce information by voice so as to be transmitted to a human.

[0005]

In order to achieve the above object, a voice synthesizing apparatus according to claim 1 of the present invention is a voice synthesizing apparatus for inputting a character code string and performing speech synthesis using the character code string. In the pronunciation information storage means for storing pronunciation information indicating the reading of a word and notation pronunciation information indicating the reading of the word, and extracting a word from the character code string to read the pronunciation of the word A reading means for selecting a reading from which the notation of the word can be understood from the notation pronunciation information and determining whether or not there is a homophone with the same pronunciation And are provided.

According to a second aspect of the present invention, there is provided the voice synthesizing apparatus according to the first aspect, wherein the pronunciation information storage means has the same phonetic variation that means that a word has a different homophone. Having notation word information, and when the reading means detects the presence of a homonym notation word in a word selected to be read based on the homophone different notation word information, performs a large vocabulary reading for the word. Is.

According to a third aspect of the present invention, in the voice synthesizing apparatus according to the first aspect, the reading means selects the reading of the word from the pronunciation information and then reads the reading of the word. It is determined whether or not a word having the same reading exists in the pronunciation information, and if a word having the same reading exists, the large vocabulary reading is performed for the word.

According to a fourth aspect of the present invention, there is provided a voice synthesizing apparatus according to any one of the first to third aspects, wherein the reading means has a first mode in which only the reading of a word is read out. A second mode for reading a large vocabulary is provided, and either one of the first mode and the second mode is selected according to a user's request.

A speech synthesizer according to a fifth aspect of the present invention is the speech synthesizer according to any one of the first to fourth aspects, further comprising reading adding means for giving a reading to the proper noun, The reading means performs large vocabulary reading for the unregistered proper noun when the reading of the proper noun selected from the pronunciation information and the reading of the proper noun given by the reading giving means are different.

An information retrieval system equipped with a speech synthesizer according to claim 6 of the present invention retrieves a database recording information and the database based on retrieval conditions and outputs a retrieval result character code string. A voice synthesizing apparatus according to any one of claims 1 to 5, further comprising: a search means; and a voice synthesizing device for inputting the search result character code sequence and performing voice synthesis according to the search result character code sequence.

An information retrieval system equipped with a voice synthesizer according to claim 7 of the present invention records information, and also records pronunciation information indicating the reading of the information and notation pronunciation information indicating the reading for understanding the notation. Inputting the search result character code string, and a search means for searching the database based on the search condition and outputting a search result character code string including a reading corresponding to the searched information and a reading for understanding the notation And a voice synthesizing device for synthesizing voice according to the search result character code string.

According to an eighth aspect of the present invention, there is provided an information retrieval system including a voice synthesizer according to the sixth or seventh aspect, wherein the database is a personal name / company name dictionary. .

According to a ninth aspect of the present invention, there is provided an information retrieval system including a voice synthesizer according to the sixth or seventh aspect, wherein the database is a place name dictionary.

[0014]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of a voice synthesizing device and an information search system including the voice synthesizing device according to the present invention will be described below in detail with reference to the drawings.

[Embodiment 1] Various forms are conceivable in a scene in which a person uses voice synthesis. For example, simple text-to-speech synthesis may be considered, but in the first embodiment of the present invention, it is assumed that a person uses the personal name database via a telephone, and the speech synthesis device and the speech synthesis device of the present invention are provided. A detailed information retrieval system will be described.

FIG. 1 is a block diagram showing the structure of an information retrieval system equipped with a speech synthesizer according to the first embodiment of the present invention. An information retrieval system equipped with the speech synthesizer shown in FIG. 1 inputs a tone signal transmitted from a telephone 10 via a telephone line, converts the tone signal into a character code string and outputs the character code string, and a search input unit 11. A person name search unit 12 as a search means of the present invention for inputting a character code string and searching a person name database 13 as a database of the present invention in which personal names are stored and outputting a search result character code string, and a search result character code string Is input, the pronunciation information dictionary 15 as the pronunciation information storage means of the present invention is searched based on the search result character code string, the pronunciation information corresponding to the search result character code string is acquired, and voice synthesis is performed. The voice synthesizer 14 as a reading means of the present invention for outputting the above information to the telephone line. In addition, personal name database 13
Can also include the company name.

The operation of the information retrieval system provided with the speech synthesizer having the above configuration will be described below.

The search input unit 11 receives the tone signal string sent from the telephone 10 via the telephone line, converts the tone signal string into a character code string, and outputs it to the person name searching unit 12.

The personal name search unit 12 operates based on the character code string input from the search input unit 11, searches the personal name from the personal name database 13, and the search result character code string shown in FIG. Output. Also, personal name search section 1
2 outputs the search request character code string shown in FIG. 2 to the voice synthesizer 14 based on the character code string input from the search input unit 11.

FIG. 2 shows an example of a search request character code string and a search result character code string used in the information search system equipped with the speech synthesizer according to the embodiment of the present invention. The personal name search unit 12 outputs the search request character code string and the search result character code string shown in FIG. 2 to the voice synthesis unit 14. The voice synthesis unit 14 performs voice synthesis according to the search request character code string and the search result character code string output from the person name search unit 12.

That is, the voice synthesizing unit 14 is the personal name searching unit 1.
Based on the character code string output from No. 2, the phonetic information dictionary 15 is used to perform voice synthesis, and the voice synthesized information is output to the telephone line.

FIG. 3 is an explanatory diagram for explaining the structure of the pronunciation information dictionary 15 used in the information search system provided with the speech synthesizer according to the embodiment of the present invention.
The pronunciation information dictionary 15 stores pronunciation information indicating reading of words. The voice synthesizer 14 receives a character code string, extracts a word from the character code string, and searches the pronunciation information dictionary 15 for pronunciation information from the notation of the extracted word. The pronunciation information is provided with a flag indicating whether or not there are homophones and different notation words for each word. When this flag is present, the speech synthesizer 14 performs large vocabulary reading described later.

As a method of checking the homophones with different pronunciations, other than setting the flag indicating the presence of the homophones with different pronunciations, after selecting the reading of the word, the pronunciation information dictionary 15 is selected using the reading. Another method is to look up the dictionary again and, if there are words with the same reading, determine that there are homophones with different phonetics.

FIG. 4 is an explanatory diagram for explaining the structure of the pronunciation information dictionary 15 used in the information retrieval system including the voice synthesizer according to the embodiment of the present invention. In addition to the pronunciation information shown in FIG. 3, the pronunciation information dictionary 15 stores notation pronunciation information that indicates the pronunciation for understanding the notation of a word. That is, the phonetic pronunciation information is information in which reading is added to each character as shown in FIG. The voice synthesizer 14 determines that if the same-sound notation flag is set for a word,
The notation pronunciation information of the pronunciation information dictionary 15 shown in FIG. 4 is drawn for each character of the notation, and after reading the word selected from the pronunciation information, a proper pose is added to add a reading that allows the notation of the word to be read. To do. As a result, the reading shown in FIG. 5 is generated.

As described above, according to the information retrieval system equipped with the speech synthesizer according to the first embodiment, both the correct reading and the reading for understanding the notation can be presented. Can be told.

It is to be noted that the voice synthesizer 14 reads a mode in which all search result character code strings are correctly read (normal output mode of search results) and a mode in which the proper noun is read based on the pronunciation information dictionary 15 (large reading mode). It is possible to provide two modes (vocabulary reading output mode).

The user can select either the normal output mode of the search result or the large vocabulary reading output mode of the search result. When the search input unit 11 receives the user's request, the mode of the voice synthesis unit 14 is changed.

As described above, by providing the voice synthesis unit with the normal output mode of the search result and the notation reading output mode of the search result, for example, when the user does not need to read the notation, the user can meet the user's request. Correspondence becomes possible.

Further, in the first embodiment, the information retrieval system provided with the speech synthesizer has been described, but it is not limited to use in the information retrieval system and the speech synthesizer can be used alone. It goes without saying that you can do it.

[Second Embodiment] In the second embodiment as well, as in the first embodiment, an information search system equipped with a voice synthesizer which is assumed to be a personal name search system via a telephone will be described.

FIG. 6 is a block diagram showing the structure of an information retrieval system including a voice synthesizer according to the second embodiment of the present invention. The information retrieval system including the voice synthesizer shown in FIG. 6 inputs a tone signal transmitted from the telephone 10 via a telephone line, converts the tone signal into a character code string and outputs the character code string, A person name search unit as a search means of the present invention, which inputs a character code string, searches the person name + pronunciation information database 50 as a database of the present invention in which the person name and its reading are stored, and outputs the search result character code string. 12 and the search result character code string are input, and the pronunciation information dictionary 15 is entered based on the search result character code string.
Voice synthesis section 1 as a voice synthesis apparatus of the present invention for searching the phonetic information, acquiring pronunciation information corresponding to the search result character code string, performing voice synthesis, and outputting the voice synthesized information to the telephone line.
4 is provided.

As shown in FIG. 7, the reading of the personal name is directly written in the personal name + pronunciation information database 50, and the reading information for which the notation is understood is also assumed assuming that the notation needs to be read. Written together. The output from the personal name search unit 12 is output in a format including a special symbol ("#" in FIG. 8) as shown in FIG. The voice synthesizing unit 14 reads the portion surrounded by the symbol "#" as it is, and converts the other portions from the character code string to reading to perform voice synthesis.

As described above, according to the information retrieval system including the voice synthesizer of the second embodiment, the notation, the correct reading of the notation, and the reading for understanding the notation are stored in the person name + pronunciation information database 50. You can read the words corresponding to the individual case of each word. Therefore, the information to be transmitted can be transmitted to humans with few errors. In addition, since the notation, the correct reading of the notation, and the reading for understanding the notation are stored in the person name + pronunciation information database 50, the structure of the pronunciation information dictionary 15 can be simplified, and a search for homophones and different notation words is required. Instead, the speed of speech synthesis processing can be increased.

[Third Embodiment] In the third embodiment, a case where the voice synthesizer according to the present invention is used as a reading device of a vehicle-mounted navigation system will be described.

FIG. 9 is a block diagram showing the structure of the speech synthesizer according to the third embodiment of the present invention. The speech synthesizer shown in FIG. 9 is connected to a vehicle-mounted navigation system 90 having map information 91 such as a CD-ROM, receives a character code string output from the vehicle-mounted navigation system 90, and based on the character code string. The pronunciation information dictionary 94 serving as the pronunciation information storage means of the present invention is searched, the pronunciation information corresponding to the character code string is acquired, the speech synthesis is performed, and the speech synthesis information of the present invention is output. The synthesizing unit 92 is provided, and the voice-synthesized information is output to the speaker 95. The voice synthesizer 92
Is provided with an unregistered word reading adding section 93 as a reading adding means of the present invention for adding reading to an unregistered character code string in the pronunciation information dictionary.

The operation of the speech synthesizer having the above configuration will be described below.

The voice synthesizer 92 receives the character code string output from the vehicle-mounted navigation system 90,
Speech synthesis is performed based on the pronunciation information dictionary 94. Since the structure of the pronunciation information dictionary 94 is the same as that described in the first embodiment with reference to FIGS. 3 and 4, the description thereof will be omitted here.

The voice synthesizing unit 92 includes an unregistered word reading imparting unit 93 as in the case of the conventional voice synthesizing unit, and separates the pronunciation information dictionary 94 for proper nouns from the input character code string. The result of adding the unregistered word reading is generated.

As a method of giving unregistered post-reading to the proper noun by the unregistered post-reading added document image 93, for example, the style of the input character code string is analyzed, and the word corresponding to the proper noun is read as a proper noun. There is a method of giving. If the word corresponding to the proper noun is composed of one Kanji character, the word is read lesson, and further two Kanji characters are used.
If it is composed of more than letters, there is a method of reading each kanji aloud and connecting the kanji readings into one proper noun reading.

FIG. 10 is an explanatory diagram for explaining the result of adding the unregistered word reading in the voice synthesizing unit 92. When the dictionary lookup result of the pronunciation information dictionary 94 and the unregistered word reading adding result by the unregistered word reading adding unit 93 are different,
As described with reference to FIGS. 4 and 5, the pronunciation reading is added after the pronunciation information dictionary lookup result.

As described above, according to the speech synthesizer according to the third embodiment, when the character code string input to the speech synthesizer 92 includes a proper noun, the proper reading and notation of the proper noun can be known. Because the reading is read aloud, it is possible to convey to humans with a few errors even if the reading is difficult to understand or if the reading is difficult to understand.

[0042]

As described above, according to the speech synthesizer of the present invention (claim 1), pronunciation information indicating the reading of a word and notation pronunciation information indicating the pronunciation of the word are stored. The pronunciation information storage means and the word is extracted from the character code string, the reading of the word is selected from the pronunciation information, and it is judged whether or not there are homophones and different notation words. Since it is equipped with a reading means that selects a reading that understands the word notation from the information and performs large vocabulary reading, it becomes possible to convey the information to be transmitted to humans in a form with few errors.

According to the speech synthesizer (claim 2) of the present invention, in the speech synthesizer according to claim 1,
The pronunciation information storage means has homophone different notation information that means that a homophone different notation word exists in the word, and the reading means, based on the homophone different notation word information, gives the homophone different notation to the word whose reading is selected. When the presence of a word is detected, we decided to perform a large vocabulary reading for the word, so it is possible to perform a large vocabulary reading with an easy method, and it is possible to convey the information to be transmitted to humans in a form with few errors. Becomes

According to the speech synthesizer of the present invention (claim 3), in the speech synthesizer of claim 1,
After the reading means selects the pronunciation of the word from the pronunciation information, it judges whether or not a word having the same reading as the pronunciation of the word exists in the pronunciation information, and if a word of the same reading exists, the word is read. Since we decided to perform a large vocabulary reading about, it is possible to perform a large vocabulary reading in an easy way, and it is possible to convey the information to be transmitted to humans in a form with few errors. Also, the configuration of the pronunciation information storage means can be simplified.

According to the speech synthesizer (claim 4) of the present invention, in the speech synthesizer according to any one of claims 1 to 3, the reading means reads out only the reading of the word. Since a mode and a second mode for reading large vocabulary are provided, and either one of the first mode and the second mode is selected according to the user's request, it is not necessary for the user to read large vocabulary. It can respond to requests and provide detailed services.

Further, according to the speech synthesizer (Claim 5) of the present invention, in the speech synthesizer according to any one of Claims 1 to 4, there is further provided a reading adding means for giving a reading to the proper noun. However, if the reading means determines that the reading of the proper noun selected from the pronunciation information differs from the reading of the proper noun given by the reading giving means, the reading of the proper noun is decided to be a large vocabulary reading. Even proper nouns and proper nouns whose pronunciation is difficult to read can be conveyed to humans with few errors.

Further, according to the information retrieval system (Claim 6) equipped with the speech synthesizer according to the present invention, the database in which information is recorded and the database are retrieved based on the retrieval condition, and the retrieval result character code string is obtained. Search method to output,
Since the voice synthesis device according to any one of claims 1 to 5 is provided, which inputs a search result character code string and performs voice synthesis according to the search result character code string. It is possible to convey to humans in a few forms.

Further, according to the information retrieval system (Claim 7) equipped with the voice synthesizer according to the present invention, the information is recorded and the pronunciation information indicating the reading of the information and the notation pronunciation information indicating the reading of the notation are provided. Input a search result character code string, and a search means that searches the database based on the search condition and outputs a search result character code string including a reading corresponding to the searched information and a reading that understands the notation Since a voice synthesizer for synthesizing voice according to the search result character code string is provided, the result of searching the database can be transmitted to humans in a form with few errors. Further, since the information is recorded in the database and the pronunciation information indicating the reading of the information and the notation pronunciation information indicating the reading for understanding the notation are recorded, the configuration of the speech synthesizer can be simplified.

According to the information search system (claim 8) equipped with the voice synthesizer according to the present invention, in the information search system according to claim 6 or 7, since the database is a personal name / company name dictionary, It is possible to convey to a human being a person / company name whose reading is difficult to understand from the notation and a person / company name whose reading is difficult to understand from the notation.

Further, according to the information search system (Claim 9) equipped with the voice synthesizer according to the present invention, in the information search system according to Claim 6 or 7, since the database is a place name dictionary, for example, in-vehicle navigation When referring to the surrounding road guidance using the output of the system, it is possible to convey a place name that is difficult to read from the notation to humans with few errors.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of an information search system including a voice synthesizer according to a first embodiment of the present invention.

FIG. 2 is an explanatory diagram showing an example of a search request character code string and a search result character code string used in the information search system including the voice synthesizer according to the first embodiment of the present invention.

FIG. 3 is an explanatory diagram for explaining a structure of a pronunciation information dictionary used in the information search system including the speech synthesizer according to the first embodiment of the present invention.

FIG. 4 is an explanatory diagram for explaining a structure of a pronunciation information dictionary used in the information search system including the voice synthesizer according to the first embodiment of the present invention.

5 is an explanatory diagram for explaining readings generated as a result of using the pronunciation information dictionaries shown in FIGS. 3 and 4. FIG.

FIG. 6 is a block diagram showing a configuration of an information search system including a voice synthesizer according to a second embodiment of the present invention.

FIG. 7 is an explanatory diagram for explaining the structure of a personal name database used in the information search system including the speech synthesizer according to the second embodiment of the present invention.

FIG. 8 is an explanatory diagram for explaining an output format of a personal name search unit in the information search system including the voice synthesizer according to the second embodiment of the present invention.

FIG. 9 is a block diagram showing a configuration of a voice synthesis device according to a third embodiment of the present invention.

FIG. 10 is an explanatory diagram for explaining a result of performing unregistered word reading addition in a voice synthesis unit in the voice synthesis device according to the third embodiment of the present invention.

[Explanation of symbols]

 10 Telephone 11 Search Input Section 12 Personal Name Search Section 13 Personal Name Database 14 Voice Synthesis Section 15 Pronunciation Information Dictionary 50 Personal Name + Pronunciation Information Database 90 Car Navigation System 91 Map Information 92 Voice Synthesis Section 93 Unregistered Word Reading Assignment Section 94 Pronunciation Information Dictionary 95 speakers

Claims (9)

[Claims] 1. In a voice synthesizer for inputting a character code string and performing speech synthesis using the character code string, pronunciation information indicating a pronunciation of a word and notation pronunciation information indicating a pronunciation for understanding the notation of the word are provided. The phonetic information storage means to be stored and a word is extracted from the character code string, the reading of the word is selected from the phonetic information, and it is determined whether or not there is a homophone different expression word. A voice synthesizing device comprising: a reading means for selecting a reading that understands the word notation from the notation pronunciation information if it exists and performing large vocabulary reading. 2. A speech synthesizer according to claim 1, wherein:
The pronunciation information storage means has homophone different notation information which means that a homophone different notation word exists in the word, and the reading means, based on the homophone different notation word information, selects the reading selected word. A voice synthesizing device characterized by performing a large vocabulary reading on the word when it is detected that the same phonetic notation word exists. 3. The speech synthesizer according to claim 1, wherein
After the reading means selects a reading of a word from the pronunciation information, it determines whether or not a word having the same reading as the reading of the word exists in the pronunciation information, and a word of the same reading exists. In the case, a speech synthesizer characterized by performing large vocabulary reading on the word. 4. The speech synthesizer according to claim 1, wherein the reading means has a first mode of reading only the reading of a word and a second mode of reading a large vocabulary, and a user The voice synthesizing device according to claim 1, wherein either one of the first mode and the second mode is selected. 5. The voice synthesizing apparatus according to claim 1, further comprising a reading adding unit that adds a reading to the proper noun, wherein the reading unit selects the proper noun from the pronunciation information. And a reading of the proper noun given by the reading adding means are different from each other, a large vocabulary reading is performed for the proper noun. 6. A database storing information, a search means for searching the database based on a search condition and outputting a search result character code string, and inputting the search result character code string, the search result character code An information retrieval system comprising a voice synthesizer, comprising: the voice synthesizer according to any one of claims 1 to 5, which performs voice synthesis according to a row. 7. A database for recording information and recording pronunciation information indicating reading of the information and notation pronunciation information indicating reading for understanding the notation, and searching the database based on a search condition, A search means for outputting a search result character code string including a corresponding reading and a reading for understanding the notation, and a speech synthesizer for inputting the search result character code string and performing speech synthesis according to the search result character code string. An information retrieval system including a voice synthesizer, comprising: 8. An information retrieval system comprising the speech synthesizer according to claim 6 or 7, wherein the database is
An information retrieval system equipped with a voice synthesizer, which is a personal name / company name dictionary. 9. An information retrieval system comprising the speech synthesizer according to claim 6, wherein the database is
An information retrieval system including a voice synthesizer, which is a place name dictionary.