JP2001005487A - Voice recognition device - Google Patents

Abstract

(57) [Summary] [PROBLEMS] To identify a speaker in a speech recognition device,
Based on the result of the determination, the quality of synthesized speech is improved by automatically setting speech synthesis parameters for determining sound quality such as a recognition response by speech synthesis. SOLUTION: When a speaker's utterance is input to a speaker discriminating unit 3, speaker discriminating information 3a to 3d stored in the speaker discriminating unit 3 in advance.
3b, a voice switching unit 4A is provided for determining the sex and age of the speaker, and automatically changing parameters for determining the sound quality such as the speech speed and pitch of the synthesized voice uttered to the speaker according to the determination result. .

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION The present invention relates to a voice recognition result and a voice guidance based on the recognition result, which are output to a speaker as synthesized voice, by changing voice synthesis parameters according to the gender and age of the speaker. The present invention relates to a speech recognition device that emits a synthesized speech that is optimal for each speaker.

[0002]

2. Description of the Related Art For example, a voice recognition device mounted on a moving body such as an automobile is a car stereo device installed in a vehicle,
2. Description of the Related Art It is used as a voice operation means of an in-vehicle device for the purpose of reducing the physical and psychological burden on a driver and operating safely when operating an in-vehicle device such as a navigation device and a telephone terminal.

[0003] The following two methods are available for the voice recognition device to notify the user of the voice recognition result or to guide the user to the next utterance in response to the utterance content of the user for voice recognition. is there. For example, a method of visually notifying by displaying characters or the like on a display panel or a display device of an in-vehicle device, or a method of audibly notifying by a synthesized voice by voice synthesis.

The above-mentioned notification to the user is performed in an operating environment such as an automobile. Therefore, from the viewpoint of contributing to the improvement of safety in driving described above, for example, the user is required to confirm the display of the recognition result. In general, it is not necessary to largely deviate the line of sight from the front of the driver's seat.

Next, the configuration of a conventional speech recognition apparatus will be described with reference to the accompanying drawings. FIG. 11 is a block diagram showing a configuration of a conventional voice recognition device. In FIG. 11, reference numeral 9 denotes a main body of a conventional voice recognition device used for voice input / output control of the navigation device 10, for example. The speech recognition device 9 includes a speech recognition unit 2 for recognizing speech content input by speech and speech information responding to the speech content based on the speech recognition result or synthesized speech information for guiding the user to the next speech content. A voice switching unit 4 provided, a voice synthesis unit 5 that performs voice synthesis based on the synthesized voice information and outputs a synthesized voice signal
It is composed of

[0006] The voice switching unit 4 provides the voice synthesis unit 5 with text data for voice synthesis Si, which is the content of voice synthesis.
a text data storage memory 4b for outputting g7, a synthesis parameter storage memory 4c for outputting a speech synthesis parameter Sig8 to the speech synthesis unit 5, a content of speech synthesis sound such as a speech recognition result response and an uttered speech guidance by a text data control signal Sig5. Text data S for speech synthesis
The voice synthesis control unit 4a instructs the text data storage memory 4b to output ig7.

[0007] As a peripheral device of the voice recognition device 9, a voice from a user is picked up and a voice voice signal Sig 1 is picked up.
Input to the voice recognition unit 2, a speech start signal Sig instructing to wait for a speech from the user
Speech switch 8 for outputting 11 to speech recognition unit 2, amplifier 6 for amplifying synthesized speech signal Sig9 output from speech synthesis unit 5 and outputting synthesized speech signal Sig10 to speaker 7, output from speech recognition unit 2. Voice recognition result signal Si
There is a navigation device 10 that performs a map search according to g2 and performs display and the like.

The voice synthesis parameter data Sig8 output from the voice synthesis parameter storage memory 4c to the voice synthesis unit 5 is a parameter for adjusting, for example, the utterance speed, pitch of the voice, and the intensity of the intonation of the synthesized voice. It is. The voice synthesis unit 5 receives voice synthesis parameter data Sig8 and text data for voice synthesis Sig7 and performs voice synthesis.
The synthesized voice signal Sig9 is output to the amplifier 6, and the synthesized voice signal Sig10 amplified by the amplifier 6 is uttered to the user as synthesized voice from the speaker 7.

Next, the function of each block will be described along the flow of signals. When the uttered voice signal Sig1 picked up by the microphone 1 is input to the voice recognition unit 2, the voice recognition unit 2 outputs a voice recognition result signal Sig2 as a voice recognition result.
Is input to the navigation device 10 and the voice synthesis control unit 4a of the voice switching unit 4.

The text-to-speech control unit 4a instructs the text data storage memory 4b to output a text-to-speech text data Sig7 relating to the content of the synthesized voice, such as a voice recognition result response and an uttered voice guidance, by the text data control signal Sig5. The text data storage memory 4b is a voice synthesis unit 5
, The voice synthesis text data Sig7, which is the content of the voice synthesis, is output.

The speech synthesis algorithm executed by the speech synthesis unit 5 generally generates a synthesized speech signal from the text data for speech synthesis Sig7. Text to
Speech).

On the other hand, a speech recognition result signal Sig2 input from the speech recognition unit 2 to the navigation device 10 is subjected to control by control means (not shown) in the navigation device 11 to execute an operation based on speech recognition. When the user operates the speech switch 8 connected to the speech recognition unit 2, the speech start signal Sig 11 is output to the speech recognition unit 2. The utterance start signal Sig11 notifies the voice recognition unit 2 of the start of the utterance voice input, and instructs to wait for the utterance.

FIG. 12 shows a configuration in which the voice recognition device 9 shown in FIG. 11 is applied to control of the navigation device 10. In this configuration, based on the output image signal Sig12 as a result of the voice recognition operation performed for the address search as a part of the map screen search operation of the navigation device 10, the address name display screen 12b of the display 12 displays [Tokyo An example is shown in which the address name of Marunouchi 1-chome, Chiyoda-ku, Tokyo] is displayed, and a map of the surrounding area is displayed on the address surrounding map screen 12a.

In FIG. 12, the synthesized speech signal Sig9 is output from the navigation device 10 for the sake of the description of the whole figure. However, as shown in FIG. 11, the synthesized speech signal Sig9 is actually output from the speech synthesis unit 5 of the speech recognition device 9. Is done.

Next, the operation of the conventional speech recognition apparatus will be described with reference to FIG.
This will be described with reference to the flowchart shown in FIG. Step S
In step S2, after the start-up of the device in 1 is started,
It monitors whether or not the user operates the utterance switch 8, and when the operation is performed, the process proceeds to step S3, and the voice recognition unit 2 is operated.
Is activated to enable speech recognition.

In step S4, the voice recognition section 2 monitors the start of the user's utterance based on the tendency of the level of the uttered voice signal Sig1 to increase, voice analysis, and the like. In S5, the voice recognition unit 2 executes voice recognition processing.

In step S13, the speech synthesis control unit 4a generates a speech synthesis word by combining the text data with the speech synthesis unit 5 based on the speech recognition result signal Sig2. The control signal Sig5 is stored in the text data storage memory 4
b.

In the last step S14, the text data Sig for speech synthesis read from the text data storage memory 4b in response to the text data control signal Sig5.
7. Based on the speech synthesis parameter Sig8 input from the speech synthesis parameter storage memory 4c, a synthesized speech signal Sig9 is generated by text-based speech synthesis.
Thereafter, the process returns to step S2 to monitor the operation of the utterance switch 8.

FIG. 14 shows a flow of a result response of the conventional speech recognition apparatus. In FIG. 14, the operation of the utterance switch 8 by the user, the utterance of the user, the response by the synthesized voice based on the recognition result of the voice recognition device 2, and the map display are shown.
An example of a map search by voice operation of the navigation device 10 will be described.

The left side of FIG. 14 shows the user's operation of the utterance switch 8 and the chronological transition of the utterance, and the right side of FIG. 14 shows a synthesized voice output which is a response of the voice recognition device 2 to the switch operation and the utterance. 7 shows mutual chronological changes such as a change in display content of the navigation device 10.

First, the user switches the utterance switch 8 at time H1.
Is operated, the voice recognition device 9 detects the beginning of the utterance at the time point K1, and transits to a state where voice recognition can be accepted. As a result, the voice recognition device 9 uses the synthesized voice K1a [Please speak voice command. ]]
Prompt the user to utter an operation word that can be recognized by voice.

The user receives the synthesized voice K1a and the uttered voice H
In step 2, utter the word "address search". The word “address search” is a voice operation command word. The command word is stored in a voice recognition dictionary stored in a memory (not shown) incorporated in the voice recognition unit 2 of the voice recognition device 9 in advance.
It is stored as reference data for a pattern matching operation in voice recognition.

After analyzing the word "address search" in the uttered voice H2 as the uttered voice signal Sig1, the voice recognizing device 9 executes a voice recognition operation by a pattern matching operation. At time K2, the uttered content is "address search".
Is obtained. Therefore, the voice recognition device 9
Is the synthesized voice K2a [Please let me know your address. ] To prompt the user to speak a specific address name.

Upon receiving the utterance guidance, the user utters "1-chome, Marunouchi, Chiyoda-ku, Tokyo" as the utterance voice H3, and the voice recognition device 9 recognizes the utterance voice H3 at time K3. , Address name based on the speech recognition result signal Sig2 [Displays around 1-chome Marunouchi, Chiyoda-ku, Tokyo. ] Is output as a synthesized speech Ka3.
Finally, at time K4, the navigation device 10 outputs the image information Sig12 to the display 12 as shown in FIG. The display 12 is [1-chome Marunouchi, Chiyoda-ku, Tokyo]
A surrounding map screen 12a and address name characters 12b are displayed.

As described above, according to the speech recognition device having the speech synthesis function, it is possible to realize a navigation device that allows the user to operate the speech recognition device by speech recognition and speech synthesis in an interactive manner.

[0026]

As described above, in a speech recognition apparatus having a speech synthesis function that uses synthesized speech for utterance guidance and speech recognition result response, it is ensured that the user can hear the synthesized speech in particular. It is indispensable to perform the operation in order to make a series of operations flow smoothly.

The conventional speech recognition apparatus uses the speech synthesis parameter condition for a fixed speech synthesis parameter setting or for a user who is difficult to hear the speech synthesis sound or wants to adjust the speech synthesis parameter. The settings were to be selectable by the user. However, such a case has the following problems.

(1) If the user does not select the condition setting of the speech synthesis parameter, the user always hears the same tone, so that the age and gender of the user are various. Given this, users will hear synthesized speech that does not match the actual state of their auditory abilities and psychologically, and the user will not have a comfortable feeling of operation, it will be difficult to hear the synthesized voice, and the voice utterance speed will be controlled. However, there is a problem that a comfortable operation feeling cannot be obtained because the tempo does not match the tempo, and furthermore, adverse effects such as inflection of the synthesized voice and incompatibility with the pitch of the voice occur.

(2) Even in the case where the user can select the condition setting of the speech synthesis parameters, there is a problem that the user who is unfamiliar with the operation of the speech recognition apparatus takes time to set the speech synthesis parameters. there were. In various running states of a vehicle, acoustic or psychological conditions affecting the user change depending on the noise environment inside and outside the vehicle and the running speed, but the user sets the speech synthesis parameters each time. It was difficult.

The present invention has been made in order to solve the above-mentioned problems, and it has been made according to the gender and age of the user of the apparatus or the noise environment surrounding the user, the content of the synthesized voice, the speech speed and the like. It is an object of the present invention to provide a speech recognition device that can automatically change parameters for determining the pitch of a voice and the like.

[0031]

According to a first aspect of the present invention, there is provided a speech recognition apparatus for collecting speech information of a speaker to be subjected to speech recognition, and a speaker physical feature characterized by the speech information. Pattern storing means for patterning and storing attributes in a plurality of types; pattern determining means for determining which of the stored attribute patterns the collected voice information includes; physical information included in the determined attribute patterns The apparatus includes sound quality determining means for determining the sound quality of the uttered sound in accordance with the attribute, and voice synthesizing means for synthesizing a voice uttered to the speaker by reflecting the determined sound quality.

The sound quality determining means of the voice recognition apparatus according to the second aspect of the present invention includes an environment detecting means for detecting an ambient environment which has a psychological effect on the speaker, and converts the detected ambient environment into a sound quality of a uttered sound. It is to reflect.

The environment detecting means of the voice recognition device according to the third aspect of the present invention is a speed detecting means for detecting a traveling speed of a moving body carrying a speaker.

[0034] The environment detecting means of the voice recognition apparatus according to the fourth aspect of the present invention is a noise detecting means for detecting noise inside and outside a moving body on which a speaker is mounted.

According to a fifth aspect of the present invention, the sound quality determining means of the voice recognition apparatus determines the sound quality of the uttered sound and the content of the utterance according to the physical attributes.

[0036]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiment 1 Hereinafter, a speech recognition apparatus according to Embodiment 1 of the present invention will be described with reference to the accompanying drawings. FIG. 1 is a block diagram showing a configuration of an in-vehicle information system in which a voice recognition device and a navigation device according to the first embodiment are combined. In the drawings, the same reference numerals as those in FIG. 11 indicate the same or corresponding parts.

In the figure, reference numeral 9A denotes a main body of the voice recognition device according to the first embodiment. This voice recognition device 9A has a microphone (voice collection means) in addition to the configuration of the conventional voice recognition device (see FIG. 10). 1) A speaker discriminating unit 3 for discriminating a male, female, or male elderly person from the uttered voice signal Sig1 input from 1 is provided.

The speaker discriminating unit 3 is classified into male voice, female voice, male elderly voice, etc., stores respective standard patterns, and outputs each pattern as phoneme pattern signals Sig3a to Sig3c. The speaker pattern memories (pattern storage means) 3a to 3c, the phoneme pattern signals Sig3a to Sig3c output from the speaker pattern memories 3a to 3c, and the speech signal Sig1 are compared. A speaker voice comparing unit (pattern determining means) 3d for determining whether the pattern is closest to the pattern by pattern matching.

Further, the speech synthesis control section (sound quality determination means) 4a according to the first embodiment receives the speaker discrimination result signal Sig4 output from the uttered speech comparison section 3d, and converts the speech synthesis parameter Sig8 into an optimal speech. Synthesis unit (speech synthesis means)
Then, the voice synthesis parameter control signal Sig6 is output to the voice synthesis parameter storage memory 4c so as to output the voice signal to the voice synthesis parameter storage signal 4g.

Next, a schematic operation of the first embodiment will be described. When the utterance voice signal Sig1 is input from the microphone 1, the speaker determination unit 3 receives the utterance voice and the speaker pattern memories 3a to 3c stored in the speaker determination unit 3 in advance.
Each of the phoneme pattern signals Sig3a to Sig3c
And an utterance voice signal Sig1 indicating the user's utterance
To which speaker pattern is closest is determined by pattern matching. As a result of this determination, the speaker determination unit 3
It is estimated that the utterer is the owner of the speech classified into which phoneme pattern.

The estimation result is output from the uttered voice comparison unit 3d to the voice synthesis control unit 4a of the voice switching unit 4 as a speaker discrimination result signal Sig4. Speaker discrimination result signal Sig4
In response, the speech synthesis control unit 4a outputs a speech synthesis parameter control signal Sig6 to the speech synthesis parameter storage memory 4c. As a result, the speech synthesis parameter storage memory 4
From c, the optimum speech synthesis parameter Sig8 of the discrimination speaker corresponding to the speaker discrimination result is output to the speech synthesis unit 5.

The detailed operation of the first embodiment will be described with reference to the flowchart of FIG. The difference between FIG. 2 and the flowchart for explaining the operation of the conventional device shown in FIG. 13 is that the optimum speech based on the speaker discrimination result between the speech recognition processing in step S5 and the speech synthesis word creation processing in step S13. The point is that the process of selecting and outputting the synthesis parameters is inserted.

The voice recognition device 9A performs voice recognition processing in step S5, and then performs speaker determination processing in step S7. As a result of the speaker discrimination, if the speaker is an adult man in step S8, the speech synthesis parameters are adjusted for an adult man in step S9. The adjustment here means that the uttered voice comparison unit 3d shown in FIG.
The speech synthesis control unit 4a outputs a speech synthesis parameter control signal Sig6 to the speech synthesis parameter storage memory 4c based on the signal, and outputs the speech synthesis parameter to the speech synthesis unit 5. This refers to an operation of designating the data Sig8.

However, as a result of the speaker determination, if the speaker is not an adult male, the process proceeds to step S10, and it is determined whether the speaker is a female. At this time, if it is determined that the speaker is a woman, parameter setting corresponding to the woman is performed in step S11. However, if it is determined that the speaker is not a female, that is, if it is difficult to determine whether the speaker is a male or a female, parameters are set for a male elderly person in step S12. This step S12 also performs the same operation as step S9.

After setting any of the above parameters, the process proceeds to step S13, where the speech synthesis control unit 4a outputs a text data control signal Sig5 to the text data storage memory 4b, and outputs the text data for speech synthesis according to the synthesized speech content to be spoken. By designating Sig7, the speech synthesis word is determined.

Thereafter, in step S14, the text data Sig7 for speech synthesis and the speech synthesis parameter data Si
Based on g8, voice synthesis is performed in the voice synthesis unit 5, and a synthesized voice signal Sig9 is output from the voice synthesis unit 5.

Next, in the first embodiment of the present invention, the situation of changing the speech synthesis parameter of the speech recognition result response based on the speaker discrimination result will be described with reference to FIG. Here, it is assumed that the progress of the user's utterance switch operation, utterance, and response voice output of the voice recognition device are the same as those in the related art shown in FIG.

In FIG. 3, the speech synthesis parameter is changed for each user, which is the speaker discrimination result, taking as an example the recognition result response speech "1 Marunouchi, Chiyoda-ku, Tokyo" in response speech K5 in FIG. The following shows the case. In the following, men under the age of 65 are referred to as adult men and those aged 65 and over are referred to as male elderly. In FIG. 3, (a) shows a voice response for an adult male, (b) shows a voice response for a female, and (c) shows a voice response for a male elderly person. Such a classification generally allows speech discrimination.

The utterance time of each voice response differs as shown in FIG. 3 such as 6 seconds, 7 seconds, and 8 seconds. This is because adult males generally tend to be superior in listening to the response voice and shorten the response voice utterance time to prefer a fast-tempo operation response as a comfortable operation environment. Conversely, consideration must be given to slowing down vocalization and making it easier to hear for elderly males. Women are almost in between.

The table shown in FIG. 3 shows each speech synthesis parameter classified for each user classification. The utterance speed, the pitch of the voice, the intensity of the intonation, the loudness of the voice, and the like, which are the respective speech synthesis parameters, are represented by qualitative expressions.

That is, user classification (standard (under 65 years old)
For each of the adult male, female, male elderly (65 years old or older), the speech synthesis parameter is set to “fast”
"Slightly slow", "Slow", if the pitch is "Slightly high", "Normal", "High", if the inflection is strong, "Normal", "Strong", "Normal", If the voice is loud, it is qualitatively expressed as “normal”, “normal”, or “loud”.

In the speaker classifications as shown in the table, it is easy for users belonging to each classification to hear and improve intelligibility, to make the response sound utterance time appropriate, and to reduce the operation tempo. It is a setting example for realizing a comfortable voice response that is preferable as an impression of hearing.

Next, a description will be given of a case in which the voice utterance synthesized based on the speaker determination result in the first embodiment is changed according to the user with reference to FIG. For adult males, only the address of "Tokyo and Chiyodakumamaru" is output, and the response voice utterance time is reduced to 6 seconds to reduce the redundancy of the response voice. Conversely, for male elderly people, the content is not shortened, and the utterance is slowed down to 8 seconds and responded with a polite utterance saying, "I will show you one of the Tokyo and Chikudakumaru." Make it easier to hear. For women, it is between the two.

As described above, in order to change the voice utterance content according to the user, the voice synthesis control section 4a
When a speaker discrimination result signal Sig4 for specifying the user is input from d, the text data for voice synthesis of the voice utterance content corresponding to the user is read out from the text data storage memory 4b. A speech synthesis word is created in consideration of parameters.

Embodiment 2 Next, a speech recognition apparatus according to Embodiment 2 will be described with reference to the accompanying drawings. FIG. 5 is a configuration diagram of the speech recognition device 9B according to the second embodiment. FIG.
Is different from the speech recognition device 9A according to the first embodiment shown in FIG.
This is to output the speed signal Sig12 output from the speed detection means 11 mounted on the vehicle to the voice synthesis control unit 4a of the voice switching unit 4B. Here, the speed detecting means 1
1 constitutes an environment detecting means.

As a result, the voice synthesis control section 4a changes the utterance time in the voice synthesis parameters according to the speed of the vehicle. As for the change of the utterance time, the parameters are quantitatively set such that the actual time is "6 seconds", "7 seconds", and "8 seconds" for each user by making the voice synthesis parameters correspond to each traveling speed.

The operation of the second embodiment of the invention will be described with reference to the flowchart of FIG. The difference between this flowchart and the flowchart for explaining the operation of the first embodiment shown in FIG. 2 is that, in the present flowchart, after the voice recognition processing in step S5, the speed signal Sig12 is fetched in step S6a, and A moving speed determination process is performed. Then, in step S7, speaker identification processing is performed.

Next, the change of the utterance content and the utterance time of the voice response for each speaker in consideration of the traveling speed of the own vehicle will be described with reference to FIG. Here, as an example, the utterance content and utterance time when the traveling speed of the vehicle is, for example, 60 km / h or less and when the traveling speed is 60 km / h or more are divided into the above-mentioned adult male, female, and male elderly persons. Shown in

In general, the higher the running speed, the faster the vocalization of the voice response, the shorter the vocalization time, and the quicker the response. This is because when the navigation device or the like is operated by the voice recognition device, the quick response to the recognition result speeds up the transition to the next operation. This is because they will be able to respond more quickly.

However, also here, it is necessary to provide a comfortable operation feeling by varying the voice response utterance time according to the age and gender of the user. For example, the utterance times of the voice response “I will show you the chief of Tokyo” are different as follows. For an adult male under the age of 65 at a running speed of 60 km / h or less, the time is 6 seconds. 5 seconds for an adult male under the age of 65 at a running speed of 60 km / h or more. If the vehicle speed is 60 km / h or less and it is for women, the time is 7 seconds.
If the running speed is 60 km / h or more and it is for women, the time is 6 seconds. 8 seconds for an adult male aged 65 or older at a running speed of 60 km / h or less. 7 seconds if the running speed is 60 km / h or more and it is for adult men over 65 years old.

Embodiment 3 Next, a speech recognition apparatus according to Embodiment 3 will be described with reference to the accompanying drawings. FIG. 8 is a configuration diagram of a speech recognition device 9C according to the third embodiment. FIG.
The difference from the voice recognition device 9A according to the first embodiment shown in FIG. 7 is that the voice recognition device 9C according to the third embodiment uses the voice switching unit 4C to synthesize the noise inside and outside the vehicle collected by the microphone 1. This is to output to the control unit 4a.
Here, the microphone 1 forms a part of the environment detecting means.

As a result, the noise level discriminating means (not shown) in the voice synthesis control section 4a detects the noise level from the input acoustic signal, and changes the utterance level in the voice synthesis parameter according to this level. . As for the change of the utterance volume, “80 dBA”, “8”
Parameters are quantitatively set, such as 5 dBA and 90 dBA.

The operation of the third embodiment of the present invention will be described with reference to the flowchart of FIG. The difference between this flowchart and the flowchart of the operation of the second embodiment shown in FIG. 6 is that, in the present flowchart, after the voice recognition processing in step S5, the in-vehicle and the acoustic signals taken in from the microphone 1 in step S6b are obtained. A process for determining the noise level outside the vehicle is performed. Then, in step S7, speaker identification processing is performed.

The speech recognition device 9C according to the third embodiment
Examples of the change of the voice response for each speaker classification include a case where the noise level in the vehicle is, for example, 75 dBA or less,
FIG. 10 shows speech synthesis parameters including a synthesized speech level (loudness) in the case of dBA or more for adult male, female, and male elderly.

For example, when the noise level is 75 dBA or less, the synthetic voice level emitted to the standard (under 65 years) adult male and female is 80 dBA, and the male elderly person (65
The synthesized voice level issued to the user is 85 dBA.

When the noise level is 75 dBA or higher, the synthesized voice level emitted to the standard (under 65 years) adult male and female is 85 dBA, and the synthesized voice level emitted to the elderly male (65 years old and over) is 90 dBA. is there. From the above, in general, it is natural that the higher the noise level is, the higher the synthetic voice level is. However, especially for elderly men, the higher the noise level becomes, the more difficult it is to hear. Therefore, the higher the synthetic voice level is, the better.

In addition, for example, when there is a noise having a large ratio of low frequency components such as a road noise or an engine sound, which is a running noise of an automobile, a higher voice tends to be easier to hear.

Embodiment 4 In the description of the first to third embodiments, for example, the speaker discriminating unit 3 is provided separately from the voice recognizing unit 2 as shown in FIG.

However, the speech recognition unit stores in advance a template of phoneme data for speech recognition for each speaker pattern (standard phoneme data to be referred to in performing speech recognition) in a memory (not shown) in the speech recognition unit. There is a so-called multi-template speech recognition method for improving the recognition rate by replacing the template based on the result of the determination and using the template for speech recognition. In such a multi-template method, the uttered voice comparison unit 3d of the speaker determination unit 3
Is included in the voice recognition unit 2.

Further, in the description of each embodiment, the user classification is classified into three groups: males aged 65 or younger, males aged 65 and over, and females. It is possible to classify.

[0071]

According to the first aspect of the present invention, a voice collecting means for collecting voice information of a speaker to be voice-recognized, and a plurality of physical attributes of the speaker characterized by the voice information are patterned. Pattern storing means for storing the extracted voice information in any of the stored attribute patterns, and pattern determining means for determining which attribute pattern is stored, and a vocal sound according to a physical attribute included in the determined attribute pattern. Since the sound quality determining means for determining the sound quality and the voice synthesizing means for synthesizing the voice uttered to the speaker by reflecting the determined sound quality are provided, the utterance that is easy for the speaker to hear and provides a comfortable operation feeling. There is an effect that the guidance voice and the voice recognition result response can be obtained by voice synthesis.

According to the second aspect of the present invention, the sound quality determining means includes the environment detecting means for detecting the surrounding environment which has a psychological effect on the speaker, and reflects the detected surrounding environment on the sound quality of the uttered sound. Thus, there is an effect that a utterance guidance voice that makes it easy for the speaker to hear and provides a comfortable operation feeling regardless of the deterioration of the surrounding environment and a voice recognition result response can be obtained by voice synthesis.

According to the third aspect of the present invention, the environment detecting means is a speed detecting means for detecting a traveling speed of a moving body on which a speaker is mounted, so that a necessary voice can be heard without losing operation timing. There is an effect that can be.

According to the fourth aspect of the present invention, the environment detecting means is a noise detecting means for detecting noise inside and outside the moving body on which the speaker is mounted, so that the voice can be recognized with high intelligibility regardless of the surrounding noise. There is an effect that can be heard.

According to the fifth aspect of the present invention, the sound quality determining means determines the sound quality and the utterance content of the uttered sound in accordance with the physical attributes, so that there is an effect that it is possible to listen to a sound without a sense of incongruity.

[Brief description of the drawings]

FIG. 1 shows a configuration of a speech recognition apparatus according to a first embodiment of the present invention;
FIG. 2 is a block diagram showing peripheral devices.

FIG. 2 is a flowchart showing an operation of the voice recognition device according to the first embodiment of the present invention;

FIG. 3 is a diagram illustrating a change in utterance conditions of a speech recognition result response for classification of a user according to the first embodiment of the speech recognition device of the present invention.

FIG. 4 is a diagram illustrating a change in utterance content of a speech recognition result response for classification of a user according to the first embodiment of the speech recognition device of the present invention;

FIG. 5 shows the configuration of a second embodiment of the speech recognition apparatus of the present invention;
FIG. 2 is a block diagram showing peripheral devices.

FIG. 6 is a flowchart showing an operation of the voice recognition device according to the second embodiment of the present invention;

FIG. 7 is a diagram illustrating a change in the utterance speed of a speech recognition result response for classification of a user according to the second embodiment of the speech recognition device of the present invention;

FIG. 8 shows a configuration of a voice recognition device according to a third embodiment of the present invention;
FIG. 2 is a block diagram showing peripheral devices.

FIG. 9 is a flowchart showing the operation of the speech recognition apparatus according to the third embodiment of the present invention.

FIG. 10 is a diagram illustrating an example of optimization of a user and a speech synthesis parameter for each noise level in Embodiment 3 of the speech recognition device of the present invention.

FIG. 11 is a block diagram illustrating a configuration of a conventional voice recognition device and peripheral devices.

FIG. 12 is a block diagram illustrating a configuration example of a navigation device to which a voice recognition device is connected.

FIG. 13 is a flowchart showing the operation of the conventional voice recognition device.

FIG. 14 is a diagram illustrating the progress of switch operation, utterance, and response voice of a conventional voice recognition device.

[Explanation of symbols]

1 microphone, 2 voice recognition unit, 3 speaker discrimination unit,
4A to 4C voice switching unit, 5 voice synthesis unit, 6 amplifier, 7 speaker, 8 utterance switch, 9A to 9C
Voice recognition device, 10 navigation device, 11 speed detection means.

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) G10L 3/00 571H

Claims (5)

[Claims] 1. A voice collecting means for collecting voice information of a speaker to be voice-recognized, a pattern storing means for patterning and storing a plurality of types of physical attributes of a speaker characterized by the voice information, Pattern determining means for determining which attribute pattern the collected voice information is included in, and sound quality determining means for determining the sound quality of the uttered sound according to the physical attribute included in the determined attribute pattern; A voice synthesizing means for synthesizing voice uttered to the speaker by reflecting the determined sound quality. 2. The apparatus according to claim 1, wherein said sound quality determining means includes environment detecting means for detecting an ambient environment that has a psychological effect on the speaker, and reflects the detected ambient environment on the sound quality of the uttered sound. Item 2. The speech recognition device according to item 1. 3. The speech recognition apparatus according to claim 2, wherein said environment detecting means is a speed detecting means for detecting a traveling speed of a moving body carrying a speaker. 4. The speech recognition apparatus according to claim 2, wherein said environment detecting means is a noise detecting means for detecting noise inside and outside a moving body on which a speaker is mounted. 5. The speech recognition apparatus according to claim 1, wherein said sound quality determination means determines the sound quality and utterance content of the uttered sound according to physical attributes.