JPH1185178A - Synthesized speech response method and its device and storage medium storing synthesized speech response program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To eliminate monotonous impressions, such as a monotone and mechan ical tones, by returning the reaction speeches selected from plural reaction speech sets along a scenario or arbitrarily assigned reaction voices. SOLUTION: A synthesized speech response sentence forming device 5 forms the output speeches of the system in accordance with the output result of a scenario interpretation apparatus 4. A reaction speech selection apparatus 6 selects the reaction speeches assigned from the reaction speech sets 9 to 12 or randomly selects the reaction speeches in accordance with the determined output of the reaction speech sets among the outputs thereof. A synthesized speech forming apparatus 7 forms the synthesized speeches outputted by an output device 8 in accordance with the instructions, questions, response sentences, etc., from the scenario interpretation apparatus 4 or the reaction speeches selected by the reaction speech selection apparatus 6. The output device 8 is a device for outputting the synthesized speeches, sounds and images, etc., and outputs and displays the state of the system and the instructions, questions and response speeches by using the same.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a synthetic voice response method and apparatus, and a storage medium storing a synthetic voice response program, and more particularly to a method of returning a reaction voice to human movement and / or operation. The present invention relates to a synthesized voice response method and apparatus for making a system user-friendly and improving the efficiency of a human interface, and a storage medium storing a synthesized voice response program.

[0002]

2. Description of the Related Art Synthetic speech is used in telephone answering devices, CD devices, ATM devices, various vending machines, calculators, talking toys, computers, and the like with the advance of technology. The synthesized voice used here may be a voice obtained by connecting pre-recorded voices (recording / editing voice synthesis), or a voice obtained by converting Japanese sentences according to rules (synthesized voice according to rules). In the prior art, these synthesized voices are uniquely determined according to the input of a human “user” using each device “system”.

[0003]

However, if the synthesized speech according to the above rules is used several times in the system, it can be expected that "when XX is performed, a synthesized speech such as XX is output". The impression is monotonous, mechanical, and monotonous. Also,
In order to solve such problems for the system, simply outputting the synthesized speech responses from the system in a random order would make it difficult for the system to have consistent features or to achieve the objectives of the system. Difficult to achieve.

[0004] For this reason, the conventional response method and apparatus based on synthesized speech give users a monotonous impression that the system is monotonous, mechanical, or monotonous. It is difficult to give the user friendliness, such as giving the user a feeling of use or having a familiarity. The present invention has been made in view of the above points, and provides a reaction voice having a certain specific feature to a user from a reaction voice set having an arbitrary feature with respect to human movement and operation. It is an object of the present invention to provide a synthesized voice response method and apparatus which can be provided to a user and a storage medium storing a synthesized voice response program.

[0005] A further object is to return a variety of reaction voices to eliminate monotonous or mechanical or monotonous impressions, and to enable a response having consistent characteristics based on a scenario. , Making the system user-friendly (e.g. fun using the system, motivated to use the system, familiarity with the system, etc.)
It is another object of the present invention to provide a synthesized voice response method and apparatus capable of improving the efficiency of a human interface and a storage medium storing a synthesized voice response program.

[0006]

FIG. 1 is a diagram for explaining the principle of the present invention. According to the present invention, in a synthesized voice response method for responding with synthesized voice based on a recognition result of an operation or an operation, an operation or an operation is recognized (step 1), and a scenario having a description including an operation of the apparatus and a response method is interpreted. (Step 2) A reaction voice is randomly selected from a plurality of reaction voice sets according to a scenario, or a reaction voice is arbitrarily designated (Step 3), and the reaction voice is returned (Step 4).

[0007] The above reaction voice set is
It is composed of a plurality of reaction voice subsets, which are a collection of reaction voices that give the user various reaction voices having specific characteristics, and each subset has a different characteristic. The above-mentioned reaction voice set is described in either a Japanese sentence, voice information expressed by phonological symbols and phonological parameters including a fundamental frequency, a duration, and power, or a voice waveform.

FIG. 2 is a diagram showing the principle of the present invention. The present invention relates to a synthesized voice response apparatus including a recognition unit that recognizes an operation or an operation, and a response unit that responds with a synthesized voice based on a recognition result by the recognition unit. The description includes an operation of the device and a response method. It has a scenario 3, a scenario interpreting unit 4 for interpreting the scenario, and a synthesized speech response sentence generating unit 5 for generating a synthesized speech based on the recognition result and the interpretation result of the scenario interpreting unit 4.

The synthesized voice response sentence generating means 5 uses a recording / editing / synthesizing device or a device for performing voice synthesis according to rules. The synthesized voice response sentence generation means 5 performs voice selection for randomly selecting a reaction voice from a plurality of reaction voice sets 61, which are a collection of reaction voices for giving various reaction voices having specific characteristics to the user. Means 51 are included.

The above-mentioned reaction voice set 61
Includes a subset of a plurality of reaction voices having different characteristics, which is a collection of reaction voices that provide a user with various reaction voices having specific characteristics. In addition, the above-described reaction voice set 61 includes voice information expressed in Japanese sentences or phonological parameters including phonological symbols and fundamental frequencies, duration, and power, or
With any of the audio waveforms.

The present invention is a storage medium storing a synthesized speech response program having a recognition process for recognizing an operation or an operation, and a response process for responding with a synthesized voice based on the recognition result by the recognition process. It has a scenario interpretation process for interpreting a scenario having a description including an operation and a response method, and a synthesized speech response sentence generation process for generating a synthesized speech based on the recognition result of the recognition process and the interpretation result of the scenario interpretation process.

[0012] The above-mentioned synthetic speech response sentence generation process includes a speech for randomly selecting a reaction speech from a plurality of reaction speech sets, which are a collection of reaction speeches for giving various reaction speeches having specific characteristics to a user. Including the selection process. According to the above, when recognizing an operation or operation and responding with a synthesized voice based on the recognition result, a scenario is interpreted, and a reaction voice randomly selected from a plurality of reaction voice sets along the scenario and It is possible to return the reaction voice arbitrarily specified.

The reaction voice set can be described in either a Japanese sentence, voice information expressed by phonological symbols and prosodic parameters such as fundamental frequency (F0), duration and power, or a voice waveform. It is possible. In addition, the reaction audio set can provide the user with specific features consistent with each reaction audio set, and can provide certain consistent and intended characteristics in the system.

[0014]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 3 shows the configuration of the voice response device according to one embodiment of the present invention. The configuration shown in the figure includes an input device 1, a sensor 2, a scenario 3, a scenario interpretation device 4, a synthesized voice response sentence generation device 5, an output device 8, and a reaction voice set 9.
-12, and a response image / response sound generation device 13.

The input device 1 is one of a pointing device such as a keyboard, a mouse, a touch panel, and a tablet, a joystick, a joypad, and other input devices. The sensor 2 includes a microphone sensor, a magnetic sensor, an optical sensor, a pressure sensor, a capacitance sensor, an ultrasonic sensor, an image recognition device, a voice recognition device, a moving image recognition device, a PB signal recognition device, and other devices for detecting physical quantities. It is assumed that

With these input device 1 and sensor 2,
Data necessary for recognizing human motion and human operation can be obtained. Scenario 3 describes the operation of the synthesized voice response device based on data detected by the input device 1 and the sensor 2. The scenario interpreting device 4 recognizes a human motion and a human operation based on data detected by the input device 1 and the sensor 2, determines an instruction, a question, a response sentence based on the scenario 4, and sets a reaction voice set. Make a decision.

The synthesized speech response sentence generation device 5 comprises a reaction speech selection device 6 and a synthesized speech generation device 7. The synthesized voice response sentence generation device 5 generates an output voice of the system based on the output result of the scenario interpretation device 4. The reaction voice selection device 6 selects a reaction voice specified from the reaction voice sets 9 to 12 or a reaction voice at random based on the output of the determination of the reaction voice set among the outputs from the scenario interpretation device 4.

The synthesized voice generation device 7 generates a synthesized voice output from the output device 8 based on an instruction, a question, a response sentence or the like from the scenario interpretation device 4 or the reaction voice selected by the reaction voice selection device 6. Generate. The output device 8 is a device that outputs synthesized speech, sound and images, and outputs and displays system status, instructions, questions, and response voices using these devices.

The reaction voice sets 9 to 12 include "indifferent feeling", "good feeling", "frank feeling",
It is configured for each reaction voice that gives the user a specific impression such as “feeling twisted”. For example, "indifferent feeling" means "Hey", "Ho",
Short words such as "Oh" and "Eh" have the characteristic that F0 is low and the change of F0 is flat, or that the change of F0 is slightly right-downward. It is a reaction voice that has the characteristics of expressing indifferentness and indifference as the meaning of words, such as "I don't care," or "Hmm."

[0020] For example, the "small impression" means that the average F0 is medium and the average speech speed is slightly longer than the average speech speed other than the reaction voice in order to give a positive and bright feeling. It is a reaction voice with a positive meaning and prosody such as "Yes", "Yes", "I see". Also, for example, the “twisted impression” has a characteristic that has a portion where F0 changes rapidly and rises to the right and has a prosody such that the average speech speed is slightly slower, It is a reaction voice that has a skeptical meaning and prosody that can be taken in a semi-negative manner, such as "should", "honto", "so-so", "maji", "so-kana". Here, the number of reaction voice sets is four, but in the voice response device, the number of reaction voice sets need not be four, but may be one or more.

The response image / response sound generation device 13 generates an output image or output sound of the system based on the output result of the scenario interpretation device 4. Here, a response method using a synthesized voice will be specifically described by taking a fortune-telling using a selection formula chart as an example. Here, the selection formula chart will be described first. The selection formula chart is "What color do you like best among the following four colors? 1, red. 2, white. 3, blue. 4,
In response to a question with two or more options, such as "green", the question and the answer are repeated, and the result of the fortune-telling is finally reached according to the answer selected. In this example, the input device 1 is a mouse or a keyboard, and is used for selecting an option. Scenario 3 describes a question sentence and options of the selection formula chart, a relationship between the selected option and the next question sentence, a result of fortune-telling, and which reaction voice set to use.

FIG. 4 shows an example of a scenario description according to an embodiment of the present invention. In the figure, the part "S1" is a table describing the correspondence between the file name of the reaction voice set used in this selection formula chart and the set number used in the scenario. The part "S2" describes the operation of the system, and describes the flow of questions, options, and results of fortune-telling in this selection formula chart.

In this selection formula chart, first, as the first question, the seventh voice registered in the reaction voice set 1 is output (S3 part). When the user selects option 1 for this question, one voice is randomly output from the reaction voice set number “3” as a response, and then the process proceeds to question number “2” as the second question ( S4 part).

By performing such repetition, the process finally proceeds to the fortune-telling result. When the user selects the option "2" after hearing the result of the fortune-telling in the part "S5", the sound file "End.pcm" is output and the processing ends. In this way, a question is presented to the user based on the description of the scenario in a synthesized voice. When the user selects an answer from the options for the question, the scenario interpretation device 4 describes whether the reaction voice set is used or not based on the description of the option and the scenario selected by the user. Decide whether to generate audio.

Further, when a reaction voice set is used, a method of determining the reaction voice set and the reaction voice (whether to present randomly or specify the reaction voice to be presented) is determined. In the case where the scenario interpreter 4 generates the speech described in the scenario 3, the synthetic speech response sentence generator 5 passes data to the synthetic speech generator 7 and uses the reaction speech set in the scenario interpreter 4. Is passed to the reaction voice selection device 6.

The reaction voice selection device 6 selects reaction voices randomly or as specified by the scenario from the reaction voice set determined by the scenario interpretation device 4, and passes the data to the synthesized voice generation device 7. In this example, there are reaction voice sets that give the user four different impressions as reaction voice sets. These four different impressions are, for example, a “good impression”, a “cold impression”, a “easy impression”,
It's something like a twisted impression.

Here, as an example of the reaction voice,
If you give a “good impression”, the change in the F0 of the voice, such as “He”, “Hoo”, “Oh”, “Eh”, which does not have a clear meaning in the words, is short. , "Characteristics" such as "upward", "downward", "shape", "shape in reverse", etc. You can express "goodness" as a word by appropriately adjusting the prosody, such as slightly increasing the average speech speed like the words "Wow,""Yeah," or changing the height of F0 greatly There is sound.

Further, when F0 greatly changes in the prosody pattern, an impression such as "a strange feeling" is given.
It is assumed that a plurality of such reaction voices are registered in the reaction voice set in advance. The registration form of the reaction voice includes a voice waveform, a phoneme symbol string and a prosody parameter, a kana accent sentence, and the like. The voice generated by the synthesized voice generation device 7 is output from the output device 8.

As described above, a question is presented in a synthetic voice based on a scenario, and a reaction voice is randomly generated from a reaction voice set that gives a consistent and specific impression to a user's answer, and is output. , Can give the user a certain consistent impression, on which the next synthesized speech to be output can be predicted, and whether the synthesized speech resulting from monotonous speech output is mechanical or monotonous Problems of the prior art such as an impression can be solved. For this reason, it is possible to make the system enjoyable to use, motivated to use, or have a familiarity.

In the present invention, among the components shown in FIG. 3, the scenario interpreting device 4, the response image / response sound generating device 13, and the synthesized voice response sentence generating device 5 are constructed as software, and the voice synthesis is performed. Disk device, floppy disk, CD-RO connected to the computer
By storing it in M or the like and executing it as necessary, general-purpose use becomes possible.

The present invention is not limited to the above-described embodiment, but can be variously modified and applied within the scope of the claims.

[0032]

As described above, according to the present invention, a human motion or operation, or an input is recognized, and a response based on the recognition result and the scenario is made by outputting a synthesized voice. By giving a specific impression to the user by a response and returning various response voices, a monotonous or mechanical or monotonous impression is eliminated.

It is possible to provide the user with user friendliness, such as having fun using the system, being interested in using the system, and being familiar with the system. This has the effect of increasing the user's sustainability for things such as educational systems and amusement devices.

[Brief description of the drawings]

FIG. 1 is a diagram for explaining the principle of the present invention.

FIG. 2 is a principle configuration diagram of the present invention.

FIG. 3 is a configuration diagram of a voice response device according to an embodiment of the present invention.

FIG. 4 shows an example of a scenario description according to an embodiment of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Input device 2 Sensor 3 Scenario 4 Scenario interpreting device, scenario interpreting means 5 Synthetic voice response sentence generating device, synthetic voice response sentence generating device 6 Reaction voice selecting device 7 Synthetic voice generating device 8 Output device 9,10,11,12 Reaction Voice set 13 Response image / response sound generation device 51 Voice selection means 61 Reaction voice set

Claims (10)

[Claims] 1. A synthetic voice response method for recognizing an operation or an operation and responding with a synthetic voice based on a result of the recognition, wherein a scenario having a description including an operation of the device and a response method is interpreted. A synthetic voice response method characterized by returning a reaction voice randomly selected from a plurality of reaction voice sets or a reaction voice specified arbitrarily. 2. The reaction voice set includes a plurality of reaction voice subsets, each of which is a group of reaction voices for giving a user various reaction voices having specific characteristics, and each subset has a different characteristic. 2. The synthetic voice response method according to 1. 3. The reaction voice set is described in any of a Japanese sentence, voice information expressed by phonological symbols and rhythm parameters including a fundamental frequency, a duration, and power, or a voice waveform. Item 1
And the synthetic voice response method according to 2. 4. A synthetic speech response apparatus comprising: recognition means for recognizing an operation or operation; and response means for responding with a synthesized voice based on a recognition result by the recognition means, the description including an operation of the apparatus and a response method. And a scenario interpreting means for interpreting the scenario; and a synthetic speech response sentence generating means for generating a synthetic speech based on the recognition result and the interpretation result of the scenario interpreting means. Answering device. 5. The synthesized speech response device according to claim 4, wherein said synthesized speech response sentence generation means uses a recording / editing / synthesis device or a device for performing speech synthesis according to rules. 6. A voice selection unit for randomly selecting a reaction voice from a plurality of reaction voice sets, each of which is a collection of reaction voices for giving a user various reaction voices having specific characteristics, said synthesized voice response sentence generation unit. The apparatus according to claim 4, further comprising means. 7. The reaction voice set includes a subset of a plurality of reaction voices having different characteristics, which is a collection of reaction voices providing a user with various reaction voices having specific characteristics.
Described synthetic voice response method. 8. The reaction voice set includes one of a Japanese sentence, voice information represented by a phonological symbol and a rhythm parameter including a fundamental frequency, a duration, and power, or a voice waveform. 8. The synthesized voice response device according to claim 7. 9. A recognition process for recognizing an operation or an operation,
A storage medium storing a synthesized voice response program having a response process for responding with a synthesized voice based on a recognition result by the recognition process, wherein a scenario interpretation process for interpreting a scenario having a description including an operation of an apparatus and a response method; A storage medium storing a synthesized speech response program, comprising: a synthesized speech response sentence generation process for generating a synthesized speech based on a recognition result of the recognition process and an interpretation result of the scenario interpretation process. 10. The synthesized voice response sentence generation process includes a voice selection for randomly selecting a reaction voice from a plurality of reaction voice sets, which are a collection of reaction voices for giving various reaction voices having specific characteristics to a user. A storage medium storing the synthesized speech response program according to claim 9 including a process.