JP2001242887A - Speech recognition device and speech recognizing navigation device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a speech recognition device which surely succeeds in speech recognition even if a portion of a word is subtly different or other pronunciation is used while users utter the same meaning contents. SOLUTION: When prescribed keywords 304 included in the recognition words of a regualr reading recognition dictionary 301, recognition words, for which the reading of the keywords 304 is made closer to actual pronunciation, are generated to produce a new recognition dictionary 302. Sound data obtained from a microphone and speech recognition data generated, based on the recognition words of the newly generated recognition dictionary, are compared to conduct speech recognition processing.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to voice recognition and a voice recognition navigation device.

[0002]

2. Description of the Related Art The present location of a car is displayed, and a wide area
2. Description of the Related Art In-vehicle navigation devices (hereinafter, referred to as navigation devices) that perform detailed display and guide a traveling direction and a remaining distance to a destination are known. Further, as one function of the navigation device, a so-called voice recognition navigation device for improving the safety of the driver by giving an operation instruction from the driver while driving is also known (for example, JP-A-09-292255).

The speech recognition software used in the speech recognition navigation device generally presses a speech switch or the like, and then calculates a correlation value between the sound data spoken by the user and the recognized word in the recognition dictionary. As a result, the recognition word having the maximum correlation value is determined as the recognition result.

[0004]

However, even if the user utters a word having the same meaning and content, if the reading of a part of the word is slightly different or another reading, There was a problem of being misrecognized.

An object of the present invention is to ensure that when a user utters a word having the same meaning, even if the reading of a part of the word is slightly different or another reading, Speech recognition device that can succeed in recognition,
It is an object of the present invention to provide a speech recognition navigation device and a recognition word generation method therefor. Another object of the present invention is to provide a recording medium on which generated data or a program for generation is recorded and a transmission medium for transmission.

[0006]

DETAILED DESCRIPTION OF THE INVENTION The present invention will be described below with reference to the embodiment shown in FIG. In order to achieve the above object, the invention according to claim 1 includes a voice input unit (201) and a storage unit (209, 210) for storing a recognized word corresponding to a word to be recognized and representing a reading of the word. , Voice input means (2
01) is applied to a voice recognition device including voice recognition processing means (208) for performing voice recognition processing by comparing voice data obtained based on recognition words with voice data obtained based on a recognition word. In (209, 210), when a part of one speech recognition target word includes a predetermined word, the reading of the predetermined word is different from the normal reading of the predetermined word. The first recognition word is stored. The invention of Claim 2 is Claim 1
In the described speech recognition device, the predetermined word is a word commonly included in a part of a plurality of words to be subjected to voice recognition, and a reading different from a regular reading of the predetermined word is obtained by reading a normal pronunciation of the word. It is made to be. Claim 3
According to the invention of the first or second aspect of the present invention, in the voice recognition device according to claim 1 or 2, a reading different from the normal reading of the predetermined word is performed only after the syllable of the Japanese syllabary in the normal reading of the predetermined word. When the syllables are arranged, the syllable "i" is replaced by the syllable "e". According to a fourth aspect of the present invention, in the voice recognition device according to the first or second aspect,
When a syllabary of "U" is arranged after the syllable of the Japanese syllabary only in the regular reading of the predetermined word, the syllable of "U" is changed to "O". This is a reading that has been replaced with syllables. According to a fifth aspect of the present invention, in the voice recognition device according to the third or fourth aspect,
The recognition word is specified by a kana including a long-sound code "-", and a syllable to be replaced is replaced with a long-sound code "-" in a reading different from a normal reading of a predetermined word. The invention of claim 6 is claim 1
In the voice recognition device described above, the storage means (209, 21)
0), a normal recognition word including a normal reading of a predetermined word and a first recognition word are stored for one speech recognition target word. According to a seventh aspect of the present invention, in the voice recognition device according to the sixth aspect, in the first recognition word, a reading different from a normal reading of a predetermined word is not read. Claim 8
The invention according to claim 6, wherein in the speech recognition apparatus according to claim 6, the first
In the recognition word, a reading different from a regular reading of a predetermined word is a reading different from the kanji when the word is represented by a kanji. According to a ninth aspect of the present invention, in the voice recognition device according to the sixth aspect, in the first recognition word, a reading different from a normal reading of a predetermined word is used as a different name or an abbreviation of the word. According to a tenth aspect of the present invention, in the voice recognition device according to any one of the first to ninth aspects, the voice recognition processing means (208) relates to one voice recognition target word including a predetermined word as a part. When a voice recognition process is performed, a first recognition word is generated and stored in the storage means (209). The invention according to claim 11 is a speech input means (201) and a storage means (209, 21) for storing a recognition word corresponding to a speech recognition target word and representing a reading of the word.
0) and voice recognition processing means (208) for performing voice recognition processing by comparing sound data obtained by the voice input means (201) with data for voice recognition generated based on the recognition word.
And a storage unit (20).
9, 210), a recognition word to which a predetermined word reading is added is stored when a predetermined word is not included in the speech recognition target word. A speech recognition navigation device according to claim 12 includes: a speech recognition device (200) according to any one of claims 1 to 11;
Map information storage means (108) for storing map information, and control means (10) for performing control for route guidance based on at least the recognition result of the voice recognition device (200) and the map information.
5). According to a thirteenth aspect, in the voice recognition processing, a voice input means (201) is provided.
The method is applied to a recognition word generation method for generating a recognition word representing a reading of a speech recognition target word for generating speech recognition data to be compared with the sound data obtained by the method. When a predetermined word is included in the section, a recognition word is generated in which the reading of the predetermined word is different from the normal reading of the predetermined word. A recording medium according to claim 14, wherein in a speech recognition process, a recognition word representing a reading of a speech recognition target word for generating speech recognition data to be compared with sound data obtained by a speech input means (201). In a recognition word generation program to be generated, when a part of one word for speech recognition includes a predetermined word, a reading of the predetermined word is different from a normal reading of the predetermined word. This is a program for recording a recognition word generation program for generating a recognition word. A recording medium according to a fifteenth aspect relates to a recognition word representing a reading of a speech recognition target word for generating speech recognition data to be compared with sound data obtained by a speech input means (201) in a speech recognition process. In the case where the data is a part of one speech recognition target word and includes a predetermined word, the recognition of the predetermined word is related to a recognized word having a reading different from the normal reading of the predetermined word. This is the recorded data. A transmission medium according to claim 16, wherein in the speech recognition processing, a recognition word representing a reading of a speech recognition target word for generating speech recognition data to be compared with the sound data obtained by the speech input means (201). In a recognition word generation program to be generated, when a part of one word for speech recognition includes a predetermined word, a reading of the predetermined word is different from a normal reading of the predetermined word. A recognition word generation program for generating the recognition word described above is transmitted. A transmission medium according to claim 17 relates to a recognition word representing a reading of a word to be subjected to speech recognition for generating speech recognition data to be compared with sound data obtained by the speech input means (201) in speech recognition processing. In the case where the data is a part of one speech recognition target word and includes a predetermined word, the recognition of the predetermined word is related to a recognized word having a reading different from the normal reading of the predetermined word. It transmits data.

In the section of the means for solving the above-mentioned problems, the description is made in correspondence with the drawings of the embodiments for easy understanding, but the present invention is not limited to the embodiments.

[0008]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS First Embodiment FIG. 1 shows a first embodiment of a vehicle-mounted navigation system according to the present invention.
It is a figure showing composition of an embodiment. The in-vehicle navigation system includes a navigation device 100 and a voice unit 200. The navigation system according to the first embodiment, when a predetermined word is included in a part of a recognized word, replaces the reading of the predetermined word with the reading that is most successfully recognized by speech, and ensures that the speech recognition It is intended to be successful.

A navigation device 100 includes a GPS receiver 101, a gyro sensor 102, and a vehicle speed sensor 10.
3, the driver 104, the CPU 105, and the RAM 10
6, ROM 107, and CD-ROM drive 108
, A display device 109, a bus line 110, and the like.

The audio unit 200 includes a microphone 201,
An A / D converter 202, a D / A converter 203, an amplifier 204, a speaker 205, a speech switch 206,
A driver 207, a CPU 208, a RAM 209,
It comprises a ROM 210, a bus line 212 and the like.
The navigation device 100 and the audio unit 200 are connected via a communication line 211.

The GPS receiver 101 has a GPS (Global P
ositioning System) Receives signals from satellites and detects the absolute position and azimuth of the vehicle. Gyro sensor 102
Is constituted by, for example, a vibrating gyroscope, and detects the yaw angular velocity of the vehicle. The vehicle speed sensor 103 detects the moving distance of the vehicle based on the number of pulses output every time the vehicle travels a predetermined distance. The two-dimensional movement of the vehicle can be detected by the gyro sensor 102 and the vehicle speed sensor 103. The driver 104 includes a GPS receiver 101, a gyro sensor 102, a vehicle speed sensor 103
Is a driver for connecting the signal from the bus line 110 to the bus line 110. That is, each sensor output is output to the CPU 1
05 is converted to readable data.

The CPU 105 controls the entire navigation device 100 by executing a program stored in the ROM 107. The RAM 106 is a volatile memory and secures a work data area. ROM 107
The nonvolatile memory stores the above-described control program and the like. The CD-ROM drive 108 uses a CD-ROM as a recording medium and stores road map information such as vector road data. The CD-ROM drive may be a DVD drive using a DVD as a recording medium or another recording device. The display device 109 displays a road map of the current location and the surrounding area of the car, route information to the destination, information of the next guidance intersection, and the like. For example, it is composed of a liquid crystal display device or a CRT. The bus line 110 is connected to the navigation device 10.
0 is a line for connecting components such as the CPU 105 via a bus.

The voice unit 200 performs voice-related processing such as voice recognition and voice synthesis. Speech switch 206
Is a switch for instructing the start of voice recognition when pressed by the user. Input of sound data is performed via the microphone 201 for a predetermined time after the utterance switch 206 is pressed. The input sound is converted into digital audio data by the A / D converter 202 and the driver 207.

In the ROM 210 of the audio unit 200,
A speech recognition software (program), a speech synthesis software (program), a speech recognition dictionary (hereinafter simply referred to as a recognition dictionary), a speech synthesis dictionary (hereinafter simply referred to as a synthesis dictionary), and the like are stored. The voice recognition software calculates a correlation value between the digital voice data and all the recognized words in the recognition dictionary, and obtains a recognized word having the highest correlation value as a recognition result. The speech synthesis software calculates data for causing a specified sentence to be uttered from a speaker. Since the contents of both softwares are publicly known, detailed description will be omitted.

[0015] The recognition dictionary is a set of data obtained by collecting a plurality of words (words) to be subjected to speech recognition. Specifically, reading data of each word specified by Hiragana, Katakana, or Romaji (actually, its character code) is stored. The words stored in the recognition dictionary are called recognized words. Each recognition word is accompanied by information such as read data, display character data of the word, and coordinate information in the case of a facility name. Details of the recognition dictionary will be described later. The synthesis dictionary stores sound source data and the like necessary for speech synthesis.

At the end of the utterance, the CPU 208
9. Execute voice recognition software using the ROM 210 or the like to perform voice recognition of digital voice data. Speech recognition software generates speech recognition data for recognition words while referring to the recognition data of the recognition words in the recognition dictionary (data specified in Hiragana, Katakana, or Romaji), and calculates the correlation value with the digital speech data I do. The correlation values with the digital voice data are calculated for all the recognized words, and the recognized words having the highest correlation value and the predetermined value or more are determined, and the speech recognition is completed. The echo-back word linked to the recognition word is converted into utterance data using speech synthesis software. Then D
Using the / A conversion unit 203, the amplifier 204, and the speaker 205, the recognition result is output as an echo back.

If any of the calculated correlation values is equal to or less than a predetermined value, it is determined that voice recognition has not been performed, and no navigation operation is performed. More specifically, a beep sound indicating recognition failure such as “puppy” is sounded, and a response (echo back) indicating “unrecognizable” is made. The bus line 212 is a bus line of the audio unit 200.

Next, the recognition dictionary will be described in detail.
FIG. 2 is a diagram showing a conventional hospital name recognition dictionary storing recognition words relating to eight hospital names. The recognition word is read data related to the facility name (the hospital name in FIG. 2). In FIG. 2, “Yomi” is described in katakana, but may be in hiragana or romaji. A character code corresponding to the specified character is stored in the recognition dictionary. Each recognition word has "display character data" and "additional information". The display character data stores a character code to be displayed for the facility name on the display device 109 or the like. In the example of FIG. 2, the kanji code of the facility name shown in the column of “display” is stored. Additional information is coordinate information on the map of the facility,
Next, various information such as the number of the recognition dictionary to be read and various attribute information of the facility are stored. In FIG. 2, only coordinate information is shown as a representative.

FIG. 3 is a diagram showing a hospital name recognition dictionary to which the present invention is applied, corresponding to the conventional hospital name recognition dictionary of FIG. The contents of “display” and “extra information” in FIG.
"Display" and "Supplementary Information"
Only "reading" is different. For example, "Oji General Hospital" will be described as an example. In FIG. 2, the reading of “Oji General Hospital” is “Ojiso Gobyo-in”, but in FIG. 3, it is “Ojiso Gobyo-in”. That is, “Sogo Byoin” has been changed to “Sogo Byoin”. The next reading of "Otaru Hospital National Sanatorium" is Fig. 2.
In FIG. 3, it is "Otarubyoukokkuritsuriyosho", but in FIG. In other words, the term “click” has been changed to “click”.

The formal reading of the above "General Hospital" is "Sougo Byoin". The legitimate reading here is
It is the original reading of the word written in a dictionary.
However, the pronunciation that is actually pronounced in daily conversations and the like is slightly different, and for example, it can be said that “Sogaw Byin” using the long note “-” is closer to the actual pronunciation. Therefore, in the first embodiment, if the words (keywords) frequently appearing in the genre are read in advance closer to the actual pronunciation, and a word including the keyword is found in the recognition dictionary. Prepares a recognized word in which the reading of the keyword is replaced with a reading close to the actual pronunciation.

FIG. 4A shows the keywords that are replaced in FIG. FIG. 4 (b)
May be replaced with a reading as shown in FIG.

FIG. 5 is a diagram showing the recognition dictionary 301 in FIG.
Is a diagram showing a configuration in which a personal computer (hereinafter, referred to as PC) 303 generates a new recognition dictionary 302 of FIG. The recognition dictionary 301 is stored on a recording medium such as a CD-ROM, and the recognition dictionary 302 is generated on a hard disk (not shown). The keyword 304 is checked in advance and stored as a file on the hard disk of the PC 303. FIG. 6 shows a flowchart of a process executed by the PC 303. Keyword 304 is a keyboard (not shown)
May be input. Generated recognition dictionary 3
02 is written in the ROM and the audio unit 200 shown in FIG.
ROM 210.

Hereinafter, the processing of FIG. 6 will be described. In step S1, one recognition word is read from the recognition dictionary file in which the recognition word of normal reading in FIG. 2 is stored. Readings closer to the actual pronunciation of words (keywords) often appearing in the genre are investigated in advance and stored in advance in a personal computer. In step S2, it is determined whether the read recognition word includes a predetermined keyword. If it is determined that the predetermined keyword is included, the process proceeds to step S3. In step S3, the pronunciation corresponding to the keyword in the recognized word is replaced with a pronunciation closer to the actual pronunciation than previously examined. On the other hand, step S
If it is determined in step 2 that neither keyword is included, step S3 is skipped and the process proceeds to step S4. In step S4, it is determined whether the processing has been completed for all the recognized words, and if not completed, the process returns to step S1 to repeat the processing. If it has been completed, the process ends.

In step S3, the recognition word of the regular reading is left as it is, and a new recognition word in which the reading corresponding to the keyword is replaced with a reading closer to the actual pronunciation than previously checked is added. Is also good.
Thereby, it is possible to perform voice recognition using both the recognized word of the regular reading and the new recognized word.

FIG. 7 is a diagram showing a flowchart of control for performing voice recognition in the voice unit 200. The control program is stored in the ROM 210 and the CPU 208
Executes the control program. This routine starts when the navigation device 100 and the audio unit 200 are turned on.

In this routine, for example, it is assumed that a voice of "facility genre name" is input, and then a "facility name" is input by voice, and a map near the relevant facility is displayed on the display device 109. In step S11, an initial recognition dictionary is prepared.
The initial recognition dictionary stores recognition words relating to facility genre names such as “hospital”, “police station”, and “post office”. In step S12, it is determined whether or not the utterance switch 206 has been pressed. If the utterance switch 206 has been pressed, the process proceeds to step S13. If not, repeat this step. After pressing the utterance switch 206, the user utters a facility genre name, for example, "hospital" within a certain period of time. Step S1
In 3, the audio signal from the microphone 201 is converted into digital audio data. In step S14, it is determined whether or not the utterance has ended. The end of the utterance is determined to be the end of the utterance when the audio signal is interrupted for a predetermined time. If it is determined that the utterance has ended, the process proceeds to step S15, and if it is determined that the utterance has not ended, the process returns to step S13.

In step S15, correlation values are calculated for the digital speech data obtained in step S13 and all recognized words in the initial recognition dictionary prepared in step S11, and the flow advances to step S16. In step S16, it is determined whether the highest correlation value among the calculated correlation values is equal to or greater than a predetermined value. If the value is equal to or more than the predetermined value, it is determined that the word has been recognized, and the process proceeds to step S17. In step S17, the recognized word having the highest correlation value is recognized as the "facility genre name", and a facility name dictionary of the genre is prepared, and echo back is output as "please see the facility name". If the genre is “hospital”, the hospital name recognition dictionary of FIG. 3 generated in advance by the routine of FIG. 6 is prepared.

On the other hand, if the highest correlation value is less than the predetermined value in step S16, it is determined that the uttered word cannot be recognized, and the process proceeds to step S23. Step S2
At 3, the voice is echoed back saying "unrecognizable" and the process returns to step S12.

When the echo back message "Please enter the facility name" is output in step S17, the user speaks, for example, the hospital name shown in FIG. In step S18, the audio signal from the microphone 201 is converted into digital audio data. In step S19, it is determined whether or not the utterance has ended. The end of the utterance is determined to be the end of the utterance when the audio signal is interrupted for a predetermined time. If it is determined that the utterance has ended, the process proceeds to step S20, and if it is determined that the utterance has not ended, the process returns to step S18.

In step S20, correlation values are calculated for the digital voice data obtained in step S18 and all the recognized words in the recognition dictionary shown in FIG. 3, and the process proceeds to step S21. As the recognition dictionary, the recognition dictionary of FIG. 3 in which the reading of the keyword is replaced in the recognition dictionary of FIG. 2 is used. Step S
At 21, it is determined whether the highest correlation value among the calculated correlation values is equal to or greater than a predetermined value. If the value is equal to or more than the predetermined value, the word is recognized, and the process proceeds to step S22.
In step S22, the recognized word having the highest correlation value is echoed back by voice.

Further, in step S22, the navigation apparatus 1 confirms that the corresponding hospital name (facility name) has been recognized.
After notifying 00, the process ends. When notifying the navigation device 100, the coordinates of the display character data and the supplementary information on the map are notified. Navigation device 100
Displays a road map near the facility on the display device 109 based on the coordinate data on the map of the hospital (facility) transmitted via the communication line 211 and the map information of the CD-ROM drive 108.

On the other hand, if the highest correlation value is less than the predetermined value in step S21, it is determined that the uttered word cannot be recognized, and the process proceeds to step S24. Step S2
In step 4, echo is returned by voice saying "unrecognizable", and the process returns to step S18.

As described above, when speech recognition is performed, a recognition dictionary storing recognition words in which keywords frequently used in a corresponding genre are replaced with pronunciations different from regular pronunciations and close to actual pronunciations is used. Like that. As a result, the speech recognition can be successfully performed. In particular, since a recognition dictionary that replaces only the reading of the keyword portion is created using a recognition dictionary that has been used in the past, it is easy and short to create a new recognition dictionary with an improved voice recognition hit rate. It can be performed in a short time and at low cost. In addition, a commercially available recognition dictionary in which regular readings are stored is purchased, and based on the purchased recognition dictionary, a recognition dictionary with an improved hit rate of voice recognition is created easily, quickly, and at low cost. You can also.

-Second Embodiment- In the first embodiment, as shown in FIGS.
The example in which the personal computer 303 generates the recognition dictionary in advance and the generated recognition dictionary is mounted in the ROM 210 of the voice unit 200 in FIG. In the second embodiment, when the corresponding recognition dictionary of FIG. 3 is used in the voice unit, the corresponding recognition dictionary (FIG. 3) is generated at that time. The configuration of the vehicle-mounted navigation system according to the second embodiment is the same as that of the first embodiment.
Since the configuration is the same as that of the embodiment shown in FIG. 1, the description is omitted.

FIG. 8 is a diagram showing a flowchart of control for performing voice recognition in the voice unit 200 according to the second embodiment. Compared to FIG. 7 of the first embodiment, step S17 is replaced with step S31,
20 is replaced by step S32, and the following description will focus on the processing of steps S31 and S32.

In the second embodiment, the audio unit 20
The ROM 210 of FIG. 2 has the conventional recognition dictionary of FIG. 2 and the read data of the keyword of the genre (regular read data and replacement read data, for example, FIG. 4A). Then, the same contents as the generation of the recognition dictionary in FIG. 6 are executed in step S31. That is, in step S31 in FIG. 8, when the recognition word having the highest correlation value is recognized as the "facility genre name", a facility name dictionary (recognition dictionary in FIG. 2) of the genre is prepared. Next, based on the prepared facility name dictionary and regular reading data and replacement reading data of the keyword (FIG. 4), the same processing as in FIG. 6 is performed to replace the reading of the keyword, and a new facility name dictionary (FIG. Is generated. Then, echo back the message "Please enter the facility name." Step S
At 32, the digital voice data acquired at step S18 and the new recognition dictionary generated at step S31 (FIG. 3)
The correlation value is calculated for all the recognized words in. Other processes are the same as those in the first embodiment, and a description thereof will be omitted.

In the second embodiment, the recognition dictionary for normal reading is stored in the ROM 210, and a new recognition dictionary is stored in the RAM 209. In this case, the RAM 209
So instead of recreating the recognition dictionary entirely,
Only a recognized word including a keyword may be generated as an additional recognized word. By doing so, it is possible to calculate the correlation value for both the regular recognition word and the additional recognition word, and it is possible to perform more accurate speech recognition.

As described above, also in the second embodiment, as in the first embodiment, it is possible to surely succeed in voice recognition. In particular, since only the keyword data needs to be stored in the ROM 210 in addition to the conventional recognition word, more accurate voice recognition can be performed with a small storage capacity.

Third Embodiment In the second embodiment, the pronunciation of words (keywords) frequently appearing in the genre closer to the actual pronunciation is checked in advance, and the read data is stored in the ROM 210. It was something to keep. However, a certain rule can be found for replacing the keyword read data shown in FIG. In the third embodiment, a new recognition word is generated using the rule. The configuration of the on-vehicle navigation system according to the third embodiment is the same as the configuration shown in FIG. 1 of the first embodiment, and a description thereof will be omitted.

First, the rules will be described with reference to "Kosei Hospital" in FIG. 4B as an example. Note that a sound indicated by one kana character is called one syllable. The official reading of "Kosei Hospital" is "Koseibyoin", and "Kooseebyoin" is shown as a replacement reading. This is because there are many people who utter (pronounce) the regular reading “Koseibyouin” as “Kooseebyouin”. As a result, the following rule is found. "E""ke"
In the case of a reading word in which "I" follows the Japanese syllabary word (syllable) such as "se", "te" or "ne", the "i" is replaced by "e"
Many people speak as if they were replaced. Also, "O"
In the case of reading words in which "U" follows the word (syllable) such as "ko", "so", "to", "no", the person who speaks as if "u" was replaced with "o" There are many. This rule is disclosed in Japanese Patent Application No. 11-255983 filed by the inventor of the present application as the inventor.

FIG. 9 is a diagram showing a flowchart of control for performing voice recognition in the voice unit 200 according to the third embodiment. Since FIG. 8 differs from FIG. 8 of the second embodiment only in that step S31 is replaced by step S41, the following description focuses on the processing in step S41.

In step S41 of FIG. 9, when the recognition word having the highest correlation value is recognized as the "facility genre name", the facility name dictionary of the corresponding genre (the recognition dictionary of FIG. 2).
Prepare Next, the processing of FIG. 10 described later is performed on the prepared facility name dictionary to obtain a new facility name dictionary (for example, FIG.
Is generated. Then, echo back the message "Please enter the facility name." The other processes are the same as those of the second embodiment, and the description is omitted.

Next, the processing of FIG. 10 will be described. FIG.
6 is a flowchart showing a control for generating a new recognition dictionary from a recognition dictionary in which conventionally used recognition words of normal reading are stored, as in FIG. 6. The difference from FIG. 6 is that the above-described rule is used.

In step S51, one recognition word is read from the recognition dictionary file in FIG. Words (keywords) that frequently appear in the genre are examined in advance and stored in the ROM 210. In this case, only the regular reading data of the keyword is stored. In step S52, it is determined whether the read recognition word includes a predetermined keyword. If it is determined that the predetermined keyword is included, the process proceeds to step S53.

In step S53, it is determined whether or not the keyword has a reading in which "I" is arranged after the Japanese syllabary word (syllable) (this is called "Edan's law"). In step S53. If it is determined that there is a rule of step, the process proceeds to step S54. In step S54, the reading "A" is replaced with "E". In step S53. If it is determined that there is no step rule, step S54 is skipped and the process proceeds to step S55.

In step S55, it is determined whether or not the keyword has a reading in which "U" is arranged after the Japanese syllabary word (syllable) (this is called "dan's rule"). In step S55. If it is determined that there is a lower rule, the process proceeds to step S56. In step S56, the reading “U” is replaced with “O”. If it is determined in step S55 that there is no step rule, step S56 is skipped and the process proceeds to step S57. In step S57, it is determined whether the processing has been completed for all the recognized words, and if not completed, the flow returns to step S51 to repeat the processing. If it has been completed, the process ends.

In the processing shown in FIG. 10, the recognition word of the regular reading is left as it is, and the recognition word of the reading replaced by the law of the spelling or the spelling of the spelling is added as a new recognition word. Is also good. Further, instead of replacing with "e" or "o", it may be replaced with a long code "-". Further, both the recognition word replaced with "e" or "o" and the recognition word replaced with long code "-" may be added.

The above is the case of the speech recognition system in which the reading is specified in hiragana or katakana. However, the same applies to the case of specifying in Roman characters. For example,
"Kosei Hospital" uses "ko
useibyouin ". "U" following "o" is replaced by "o"
Replace "i" following "e" with "e" to replace "k"
A new recognition word "ooseebyooin" is generated.

As described above, if the vowel continues with "A" in the regular reading keyword, it is replaced with "Ae" or "A", and if the vowel follows "Oh", "Oh" or "O". Was replaced with a reading. As a result, a recognized word close to the actual utterance is prepared,
The probability of successful speech recognition increases.

Note that, in the third embodiment, the rule of the dan and the rule of the dan are not applied to the entire recognized word, but are applied only to a predetermined keyword. The reason will be described.
For example, assume that the place name "Wakae Iwata" has a hospital name of "Wakae Iwata Central Hospital". The regular reading in this case is "Wakaei Wachuobouin". When the entire recognition word is read by applying the rules of the stanchions and the rules of the stanchions, the result is "Wakae-e-watachiooobyoin". When the prolonged code is applied, the result is “Wakae Watachu Obyoin”. However, "Wakae Iwata" is not pronounced as "Wakae Ewata" or "Waka Ewata".

As described above, if the rule of the column and the rule of the column are applied to the entire recognition word, even the reading which should not be replaced is replaced, and conversely, the speech recognition becomes low in accuracy. In the third embodiment, in order to avoid such a problem, when there is a predetermined keyword, only the keyword is replaced. As a result, it is possible to automatically generate reading replacement words (paraphrase words, new recognition words) by applying the rules of the dan and the rules of the dan, and ensure that the reading that you do not want to replace is not replaced. Can be. As a result, a highly accurate replacement word can be automatically generated.

In the third embodiment, as in the second embodiment, an example has been described in which a new recognition dictionary is generated at the time of performing speech recognition of the genre. However, similarly to the first embodiment, a new recognition dictionary to which the rule of the stage and the rule of the stage are applied may be generated in advance by the personal computer 303 (FIG. 5). As a result, the same effect as that of the first embodiment can be obtained, and at the same time, it is possible to avoid the replacement of the reading that should not be replaced.

Fourth Embodiment In the first embodiment, an example has been described in which a new recognized word is generated by replacing a predetermined keyword with a pronunciation that is closer to the actual pronunciation. In the fourth embodiment, several readings of a predetermined keyword are prepared, and a new recognition word including each reading of the keyword is added to the recognition dictionary. The configuration of the in-vehicle navigation system according to the fourth embodiment is the same as the configuration of FIG. 1 according to the first embodiment, and a description thereof will be omitted.

A recognition dictionary according to the fourth embodiment will be described. FIG. 11 is a diagram showing a conventional recognition dictionary in which genres store recognition words relating to post offices, government offices, and inter lamps. FIG. 12 is a diagram showing a recognition dictionary corresponding to FIG. 11 to which the present invention is applied in the fourth embodiment.

The case where the facility name "Zama Post Office" is designated when the genre in FIG. 11A is "Post Office" will be described. The navigation system prompts you to select a genre name, and utters "Yubinkoku". After that, when the user is prompted to "Please enter the facility name", there are times when he is confused whether to speak "Zama Yubinkyo" or only "Zama". In the recognition dictionary of FIG.
For the "Zama Post Office", two reading recognition words "Zama Yubinkyo" and "Zama" are stored. Therefore, both "Zamayu Binkyo" and "Zama" can be recognized successfully.

Next, the genre in FIG.
The case where the facility name "Aikawa Town Office" is designated at the time of will be described. When prompted by the navigation system, "Please select the facility name," in addition to whether to say "Aikawa Machi and Kakuba" or "Aikawa Machi" and "Aikawa Choyaku""". This is because the kanji character "town" may be called "machi" or "cho" depending on the region. In the recognition dictionary of FIG. 12B, “Aikawa Machi Yakuba” and “Aikawa Machi”
Since the three recognition words "Aikawachoyakuba" are prepared, the voice recognition succeeds no matter how it is uttered. The same applies to the “village office”.

That is, in FIG. 12, when the last word of the recognition word is a predetermined keyword, a new recognition word replaced by a plurality of readings of the predetermined keyword is added. FIG. 13 is a diagram in which readings replaced with the keywords in FIG. 12 are arranged. For example,
If the recognition word is "Aikawa Machiyakuba",
When searching for the keyword in FIG. 13 in the reading, and finding the keyword “Machiyakuba”,
Recognition words replaced by the replacement reading of the keyword, that is, "Aikawamachi" and "Aikawachoyakuba" are added.

The reading to be replaced includes a case where the reading of the keyword is omitted, as shown in “Yubinkyo” and “Intarjunji” in FIG. In addition, the reason why "intar-chunji" is replaced with the pronunciation of "ai-shi" or "intar" is because it is generally used as another name of "interchange". It is because it is another name to replace "Sabisu Eria" with "Peking". It can be said that "pink king" is abbreviated as "pink king". FIG.
Is a name composed of a geographical name and a facility name. Therefore, the keyword can be the facility name of the genre.

FIG. 14 is a diagram of FIG. 12 obtained from the recognition dictionary of FIG.
3 is a flowchart of control for generating a recognition dictionary for the "." The generation of the recognition dictionary in FIG. 12 is executed by the personal computer 303 as in FIG. 5, and the generated recognition dictionary is RO
ROM2 of the audio unit 200 of FIG.
10 mounted.

In step S61, one recognition word is read from the recognition dictionary file in which the recognition words of normal reading in FIG. 11 are stored. As shown in FIG. 13, replacement reading of words (keywords) frequently appearing in the genre is investigated in advance and stored in the personal computer 303 in advance. In step S62, it is determined whether the read recognition word includes a predetermined keyword. If it is determined that the predetermined keyword is included, step S62 is performed.
Go to 63. In step S63, a new recognized word is generated in which the recognized word is replaced with a reading corresponding to the keyword. On the other hand, if it is determined in step S62 that no keyword is included, step S63 is skipped and the process proceeds to step S64. In step S64, it is determined whether the processing has been completed for all the recognized words, and if not completed, the process returns to step S61 to repeat the processing.
If it has been completed, the process ends.

In the fourth embodiment, the recognition dictionary of FIG. 12 is generated in advance by the personal computer 303 of FIG. 5, and the generated recognition dictionary is generated by the voice unit 20 of FIG.
The example shown in FIG. But the second
In the same manner as in the embodiment, when the corresponding recognition dictionary is used in the voice unit, the corresponding recognition dictionary is stored in the RAM 20.
9 may be generated. In this case, FIG.
Since only the recognition dictionary and the replacement data of the keywords in FIG. 13 need to be stored in the ROM 210, the capacity of the ROM 210 is reduced, leading to cost reduction.

In the above-described fourth embodiment, when the recognition word includes a predetermined keyword, a recognition word in which a variation in the reading of the keyword is taken into account is added, so that the voice recognition is surely successful. be able to.

The contents of the fourth embodiment can be combined with the contents of the first to third embodiments. For example, consider a case where the first embodiment and the fourth embodiment are combined. "Geographic name" in FIG.
Of the words "+ facility name", the geographical name is the target keyword of the first embodiment, and the facility name is the target keyword of the fourth embodiment. For example, “Dangozaka → Dangozaka” is added to the target keyword in the first embodiment. As a result, “rigging hill S” in FIG.
"A" is a regular reading "Dango Uzakasa Biseria"
However, new recognition words are generated for "Dango-za-ka-e-seria", "dango-za-ka-e-sue", "dango-za-ka-king", and "dango-za-ka".

-Fifth Embodiment- In the fourth embodiment, as shown in FIG. 12A, for example, when a predetermined keyword is included in a recognition word, a recognition word obtained by deleting the keyword is added. An example of doing this has been described. In the fifth embodiment, a keyword (genre name or the like) that is not included in the recognized word is added. For example, if the genre is "station", the recognition words in the station recognition dictionary will usually end with "eki"
It is the one with the reading. However, if the “Eki” is not included in the recognition word, a recognition word with “Eki” added is added. The configuration of the vehicle-mounted navigation system according to the fifth embodiment is the same as that of the first embodiment shown in FIG.
Since the configuration is the same as that described above, the description is omitted.

For example, in a recognition dictionary whose genre is “station”,
If there is a recognition word “Tokyo”, a new recognition word “Tokyo Eki” is added. In particular,
The ROM 210 has keyword data to be added corresponding to a certain genre, and a recognition word is searched. If the keyword is not included, a recognition word including the keyword is added. This keyword may be plural. When there are a plurality of keywords, a recognition word to which each keyword not included is added may be added.
In this manner, when the genre is "station", speech recognition succeeds even if "Tokyo" is uttered or "Tokyo eki" is uttered.

In the first to fifth embodiments, the on-vehicle navigation system has been described. However, the present invention is not limited to this. The present invention can be applied not only to a vehicle-mounted device but also to a portable navigation device. Furthermore, the present invention is applicable not only to navigation devices but also to all devices that perform voice recognition.

In the first to fifth embodiments, the configuration in which the navigation device 100 and the audio unit 200 are separated has been described. However, the present invention is not limited to this. It may be configured as one navigation device including a sound unit inside. Further, the control program, the recognition dictionary, and the like can be provided on a recording medium such as a CD-ROM. Furthermore, it is also possible to provide a control program, a recognition dictionary, and the like on a recording medium such as a CD-ROM, and realize the system on a computer such as a personal computer or a workstation. Further, a control program, a recognition dictionary, and the like can be provided through a telecommunication line (transmission medium) represented by the Internet.

In the first to fifth embodiments, when a search for a facility name is successful in the voice unit 200, the content is notified to the navigation apparatus 100, and the navigation apparatus 100 performs one of the navigation processes such as road guidance. Although an example of displaying a map near the facility has been described, the present invention is not limited to this example. Navigation device 10
In the case of 0, a route search, a route guidance and other various navigation processes can be considered based on the result of a successful search by the voice unit 200.

[0069]

Since the present invention is configured as described above, it has the following effects. According to the first to tenth aspects of the present invention, when performing speech recognition, for example, a recognition dictionary storing recognition words in which keywords frequently used in a corresponding genre are replaced with readings different from regular readings can be used. Therefore, it is possible to surely succeed in the speech recognition. In particular, since it is possible to create a recognition dictionary that replaces only the reading of the keyword part using a recognition dictionary that has been used conventionally or a commercially available recognition dictionary, a new speech recognition hit rate has been improved. A recognition dictionary can be easily created in a short time and at low cost, and the cost of an apparatus with an improved speech recognition rate can be reduced. In particular, the invention of claim 2
Since the pronunciation to be replaced is the pronunciation of normal pronunciation, the speech recognition rate is further improved. According to the third to fifth aspects of the present invention, reading replacement is automatically performed based on a certain rule, so that it is not necessary to previously store reading replacement data, which leads to a reduction in memory capacity. Also, since only readings of predetermined words (keywords) are replaced, readings that should not be replaced can be surely not replaced. That is, it is possible to surely avoid the problem of replacing readings that are originally not desired to be replaced due to automatic replacement based on a certain rule. In the invention of claims 6 to 9, in addition to the reading of a predetermined word (keyword) determined in advance, the pronunciation of the word in a different kanji or replaced with another name or abbreviation is also recognized as a recognition word. Since it is stored, even if the user utters in various variations, the voice recognition is successfully achieved. According to the tenth aspect of the present invention, the first recognition word is generated at the time when the speech recognition process is actually performed. It is stored, and a new recognized word can be developed on the RAM based on the contents. Thereby, the storage capacity of the ROM can be reduced. On the contrary,
Although the storage capacity of the RAM is increased, it is not a problem because the RAM is overwritten as needed and used as a work area. According to the eleventh aspect, even when a predetermined word (keyword) is added and uttered, the speech recognition can be successfully completed. According to the twelfth aspect of the present invention, the above effects can be obtained in the voice recognition navigation device. Claim 1
According to the invention of the third aspect, it is possible to create a recognition dictionary in which only the reading of the keyword portion is replaced by using a recognition dictionary conventionally used or a commercially available recognition dictionary. A simple recognition dictionary can be created easily, in a short time and at low cost.

[Brief description of the drawings]

FIG. 1 is a diagram showing a configuration of an on-vehicle navigation system of the present invention.

FIG. 2 is a diagram showing a recognition dictionary for normal reading in the first embodiment.

FIG. 3 is a diagram illustrating a new recognition dictionary according to the first embodiment.

FIG. 4 is a diagram showing a list of keyword and replacement readings according to the first embodiment.

FIG. 5 is a diagram illustrating a configuration for generating a new recognition dictionary from a recognition dictionary of normal reading in the first embodiment.

FIG. 6 is a diagram illustrating a flowchart of a process of generating a new recognition dictionary from a recognition dictionary of normal reading in the first embodiment.

FIG. 7 is a diagram illustrating a flowchart of control for performing voice recognition according to the first embodiment.

FIG. 8 is a diagram illustrating a flowchart of control for performing voice recognition according to the second embodiment.

FIG. 9 is a diagram illustrating a flowchart of control for performing voice recognition according to the third embodiment.

FIG. 10 is a diagram illustrating a flowchart of a process of generating a new recognition dictionary from a recognition dictionary of normal reading in the third embodiment.

FIG. 11 is a diagram showing a recognition dictionary for normal reading in the fourth embodiment.

FIG. 12 is a diagram illustrating a new recognition dictionary according to the fourth embodiment.

FIG. 13 is a diagram showing a list of keyword and replacement readings according to the fourth embodiment.

FIG. 14 is a diagram illustrating a flowchart of a process of generating a new recognition dictionary from a recognition dictionary of normal reading in the fourth embodiment.

[Explanation of symbols]

 REFERENCE SIGNS LIST 100 Navigation device 101 GPS receiver 102 Gyro sensor 103 Vehicle speed sensor 104 Driver 105 CPU 106 RAM 107 ROM 108 CD-ROM drive 109 Display device 110 Bus line 200 Audio unit 201 Microphone 202 A / D conversion unit 203 D / A conversion unit 204 Amplifier 205 Speaker 206 Speaking switch 207 Driver 208 CPU 209 RAM 210 ROM 211 Communication line 212 Bus line 301 Regular reading recognition dictionary 302 New recognition dictionary 303 Personal computer 304 Keywords

Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat II (reference) G10L 15/28 G10L 3/00 561H

Claims (17)

[Claims] 1. A voice input means, a storage means for storing a recognition word corresponding to a word to be recognized and representing a reading of the word, a generation based on the sound data obtained by the voice input means and the recognition word And a voice recognition processing unit that performs a voice recognition process by comparing the obtained voice recognition data with the obtained voice recognition data. A first recognition word in which the reading of the predetermined word is different from the normal reading of the predetermined word when the first word is included. 2. The speech recognition device according to claim 1, wherein the predetermined word is a word commonly included in a part of a plurality of words to be recognized, and is different from a regular reading of the predetermined word. The speech recognition device is characterized in that the pronunciation is a pronunciation that is normally pronounced for the word. 3. The speech recognition device according to claim 1, wherein a reading different from the normal reading of the predetermined word is performed only after the normal reading of the predetermined word after the syllable of the Japanese syllabary. A voice recognition device characterized in that when syllables of "i" are arranged, the syllable of "i" is replaced by a syllable of "e". 4. The voice recognition device according to claim 1, wherein a reading different from the normal reading of the predetermined word is performed only after the syllable of the Japanese syllabary in the normal reading of the predetermined word. A voice recognition device characterized in that when syllables of "U" are arranged, the syllables of "U" are replaced with syllables of "O". 5. The speech recognition device according to claim 3, wherein the recognition word is specified by a kana including a long code "-", and in a reading different from a normal reading of the predetermined word,
A speech recognition apparatus, wherein the syllable to be replaced is replaced with a long code "-". 6. The speech recognition apparatus according to claim 1, wherein the storage means stores a regular recognition word including a regular reading of the predetermined word with respect to the one speech recognition target word. A speech recognition device, wherein the speech recognition device is stored together with one recognition word. 7. The speech recognition device according to claim 6, wherein in the first recognition word, a reading different from a normal reading of the predetermined word is a reading of the word. Speech recognition device. 8. The speech recognition apparatus according to claim 6, wherein, in the first recognition word, a reading different from a normal reading of the predetermined word is a reading different from the normal reading of the kanji when the word is represented by a kanji. A speech recognition device, characterized in that: 9. The speech recognition apparatus according to claim 6, wherein in the first recognition word, a reading different from a normal reading of the predetermined word is a different name or an abbreviation of the word. Voice recognition device. 10. The speech recognition device according to claim 1, wherein the speech recognition processing unit is configured to associate the speech recognition target word with a part of the predetermined word. A speech recognition apparatus, further comprising: a generation unit configured to generate the first recognition word and store the generated first recognition word in the storage unit when performing the voice recognition process. 11. A speech input means, a storage means for storing a recognition word corresponding to a speech recognition target word and representing a reading of the word, a generation based on sound data obtained by the speech input means and the recognition word. A voice recognition processing unit that performs a voice recognition process by comparing the obtained voice recognition data with the obtained voice recognition data, wherein the storage unit includes a predetermined word as the voice recognition target word. If not, the speech recognition device stores a recognition word to which the reading of the predetermined word is added. 12. A voice recognition device according to claim 1, a map information storage means for storing map information, and a road based on at least a recognition result of said voice recognition device and said map information. A voice recognition navigation device, comprising: control means for performing control for guidance. 13. A recognition word generation method for generating a recognition word representing a reading of a word to be voice recognition for generating voice recognition data to be compared with sound data obtained by voice input means in a voice recognition process. And generating a recognized word in which the reading of the predetermined word is different from the normal reading of the predetermined word when a part of one word to be recognized includes a predetermined word. A recognition word generation method. 14. A recognition word generation program for generating a recognition word representing a reading of a word to be voice recognition for generating voice recognition data to be compared with sound data obtained by voice input means in a voice recognition process. When a part of one target speech recognition word includes a predetermined word, a recognition word is generated in which the reading of the predetermined word is different from the normal reading of the predetermined word. A recording medium characterized by recording a recognition word generation program. 15. Data relating to a recognized word representing a reading of a word to be subjected to voice recognition for generating voice recognition data to be compared with sound data obtained by voice input means in voice recognition processing, wherein In the case where a part of one of the words to be recognized includes a predetermined word, the data of the recognized word having the reading of the predetermined word different from the normal reading of the predetermined word is recorded. Characteristic recording medium. 16. A recognition word generating program for generating a recognition word representing a reading of a word to be voice recognition for generating voice recognition data to be compared with sound data obtained by voice input means in a voice recognition process. When a part of one target speech recognition word includes a predetermined word, a recognition word is generated in which the reading of the predetermined word is different from the normal reading of the predetermined word. A transmission medium characterized by transmitting a recognition word generation program. 17. A speech recognition process for generating speech recognition data to be compared with sound data obtained by speech input means, the speech recognition data representing recognition words representing readings of words to be recognized. In the case where a part of one of the words to be recognized includes a predetermined word, it is necessary to transmit data on the recognized word in which the reading of the predetermined word is different from the normal reading of the predetermined word. Characteristic transmission medium.