JP3805065B2 - In-car speech synthesizer - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an in-vehicle speech synthesizer that synthesizes character string data received on the side of a traveling vehicle from an information center side into a speech by fitting it into a prosodic pattern. Related to reducing the burden on the speech synthesizer.
[0002]
[Prior art]
FIG. 4 is a diagram showing a conventional on-vehicle speech synthesizer. As shown in this figure, on the information center side, the data generated by the inset synthesized data generation unit 1 is modulated and transmitted to the in-vehicle terminal side via the transmitter 2. The transmission data is composed of “sentence number + character string (word reading + accent)”. Here, the character string (word reading + accent) is specifically configured as “word 1 (word reading + accent) + word 2 (word reading + accent) +... + Word n (word reading + accent)”. Furthermore, the sentence number is a number transmitted to represent information of the prosodic pattern that gives the intonation to the character string. In other words, the inset composite data generation unit 1 on the information center side adds a plurality of words (word readings and accents) constituting a character string to the prosodic pattern in order to reduce the burden of language processing on the destination in-vehicle terminal side. Transmission data that can be freely replaced and fitted is generated.
[0003]
On the other hand, on the in-vehicle terminal side, the transmission data is received by the receiver 3 mounted on the vehicle, demodulated, and temporarily stored in the input buffer 3A. The prosodic pattern storage unit 4 stores a plurality of prosodic patterns, inputs “sentence number” in the input data, and outputs a prosodic pattern corresponding to the sentence number. The prosody control unit 5 inputs “character string (word reading + accent)” in the input data, inputs the prosodic pattern from the prosody pattern storage unit 4, and word 1 (word reading + accent) + word 2 (word reading + Prosody control is performed in which an accent) +... + Word n (word reading + accent) is inserted into the prosodic pattern.
[0004]
Note that the accent is a pitch pattern specific to the word (a temporal change in pitch frequency), and the intonation is the pitch pattern in the expiratory paragraph.
FIG. 5 is a diagram showing an example of a prosodic pattern stored in the prosodic pattern storage unit 4 of FIG. As shown in FIG. A prosodic pattern A having two intonations is stored for one “Today is (word 1), sunny (word 2)”. The prosody control unit 5 performs prosody control in which the words 1 and 2 are inserted into the prosody pattern A.
[0005]
Example sentence No. For 2, “Today is (word 1), clear in the morning (word 3)”, a prosodic pattern B having two intonations different from the above is stored. The prosody control unit 5 performs prosody control to insert the words 1 and 3 into the prosody pattern B.
Returning to FIG. 4, the phoneme segment dictionary unit 6 registers a plurality of speech waveform segments having a constant pitch frequency obtained from natural speech in the waveform dictionary.
[0006]
The speech generation processing unit 7 selects a speech waveform from the phoneme segment dictionary unit 6 for each phoneme of the character string generated by the prosody control unit 5, and superimposes the speech waveform on the pitch pattern of the accent and prosody pattern. , Repeat or thin out to synthesize voice and output to speaker.
[0007]
[Problems to be solved by the invention]
However, in the in-vehicle speech synthesizer, for example, as shown in the lower part of FIG. 5, an additional character string “Today is (word 1), clear in the morning (word 3), and cloudy in the afternoon (word 3). ”Is transmitted, and when the prosodic pattern storage unit 4 does not store prosodic patterns having three different intonations as in the case of such a sentence example, it is necessary to store a new prosodic pattern. . This is because if there is no new prosodic pattern, for example, if prosodic pattern B is used, the portion “cloudy in the afternoon (word 3)” is synthesized without speech intonation, resulting in a decrease in sound quality. Since the speech synthesizer is used for many vehicles on the in-vehicle terminal side, if there is frequent text data addition or data change on the information center side, a new prosodic pattern will be created for all vehicles on the in-vehicle terminal side. Although it must be frequently stored in the prosodic pattern storage unit 4, there is a problem that this is practically impossible to process.
[0008]
Therefore, in view of the above problems, the present invention makes it possible to synthesize speech without requiring new prosody pattern data addition or change processing on the in-vehicle terminal side even if transmission data is arbitrarily changed on the information center side. An object of the present invention is to provide an in-vehicle speech synthesizer.
[0009]
[Means for Solving the Problems]
In order to solve the above problems, the present invention provides an in-vehicle speech synthesizer that synthesizes character string data transmitted from an information center side to an in-vehicle terminal side into speech, and the character string data is added to the information center side. reading a plurality of words constituting and with said word accents, and inset composite data generator for generating a temporal change information der Ru prosody pattern of frequencies in breath, generated by the fitting combined data generating unit A transmission unit that transmits character string data including word reading and accent and prosody pattern data including prosody pattern, and the character string data transmitted from the information center side to the in-vehicle terminal side and a receiver for receiving said prosodic pattern data, received by the receiving unit, reading of the pre Kitan words in the prosodic pattern and A prosody control unit for performing prosody control by fitting the Kia accent, characterized in that is provided. By this means, it is not necessary to add or improve the text to be inserted by the speech synthesizer on the in-vehicle terminal side, and can be added and improved easily because it can be performed collectively on the information center side.
[0010]
The character string data is divided into exhalation paragraphs to perform speech synthesis, and the inset synthesis data generation unit transmits a control command for controlling whether transmission data is transmitted in sentence units or in exhalation paragraph units. In addition, the prosody control unit performs prosody control according to the received control command . By this means, it is possible to efficiently transmit data for speech synthesis between communication data other than speech synthesis.
[0011]
The prosody control unit performs prosodic control at the end of reception of the prosody pattern received for each exhalation paragraph, data on the reading of the plurality of words, and data on the accent of the word temporarily stored, and performs speech synthesis Start processing. By this means, the prosody control is performed on the temporarily stored data, so that the input buffer capacity on the in-vehicle terminal side can be reduced, and the synthesized sound can be generated without interruption.
[0012]
A voice type code and a prosodic parameter corresponding to the voice type code are further transmitted to the character string data transmitted from the information center side, and the in-vehicle terminal side corresponds to the voice type code and the prosodic parameter. Synthesize the voice to be played. By this means, it is possible to perform speech synthesis with the intended speech type on the information center side.
The vehicle type terminal side requests the information center side to select a voice type. By this means, it is possible to perform speech synthesis with the intended speech type on the in-vehicle terminal side.
[0013]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments of the present invention will be described below with reference to the drawings.
FIG. 1 is a diagram for explaining an example of an in-vehicle speech synthesizer according to the present invention, and FIG. 2 is a diagram for explaining an example of transmission data from the information center side. As shown in FIG. 1, on the information center side, the inset synthesized data generation unit 10 analyzes input information, which is a character string of a kanji-kana mixed sentence, divides it into words, reads a word, adds accents, and generates a phonetic symbol string An information analysis unit 11 for converting into grammatical information, reading and accent information to the information analysis unit 11, and a prosody storage unit 13 for storing a prosodic pattern which is intonation for the input information in advance. a prosodic pattern selection unit 14 that selects from the prosodic storage unit 13 prosody pattern corresponding to the input information is converted by the information analysis unit 11, prosodic pattern and information analysis selected by prosodic pattern selection unit 14 to the input information A transmission data generation unit 15 is provided that generates transmission data from the character string converted into a phonetic symbol string obtained by the unit 11 . The data generated by the transmission data generation unit 15 is modulated by the transmitter 20 and transmitted to the receiver 30. A prosodic pattern stored in the prosodic storage unit 13 is newly added in response to a change in input information, and maintenance of the prosodic pattern can be performed.
[0014]
FIG. 2 is a diagram illustrating an example of transmission data from the information center side. As shown in the figure, the transmission data generation unit 15 generates data such as prosodic pattern + word 1 (word reading + accent) + word 2 (word reading + accent) +... Word n (word reading + accent). To do.
On the other hand, on the in-vehicle terminal side, the transmission data is received by the receiver 30 mounted on the vehicle, demodulated, and temporarily stored in the input buffer 30A. The prosody control unit 50 generates a synthesized character string that synthesizes the character string data of each unit into a continuous character string as a sentence, and a data separation unit 51 that separates the received information signal into prosodic pattern data and character string data. A prosody reproducing unit 53 that reproduces each separated prosodic pattern data, and a fitting unit that inserts the synthesized character string synthesized by the synthesized character string generating unit 52 into the prosodic pattern data reproduced by the prosody reproducing unit 53 54.
[0015]
The phoneme segment dictionary unit 60 registers a plurality of speech waveform segments having a constant pitch frequency obtained from male and female natural speech in the waveform dictionary. The voice generation processing unit 7 has the same configuration as that shown in FIG.
In this way, the information center side transmits the prosodic pattern together with word reading and accented character string data, and the in-vehicle terminal side synthesizes speech using this prosodic pattern. Even if a sentence is arbitrarily added or changed, the in-vehicle terminal side does not need any processing for adding or changing a prosodic pattern as in the prior art, and can easily add or improve a sentence.
[0016]
FIG. 3 is a diagram for explaining another example of transmission data from the information center side. As shown in this figure, the inset synthesized data generation unit 10 divides one sentence into exhalation paragraph units each composed of a plurality of words, provides a prosodic pattern for each exhalation paragraph unit, and transmits it for each exhalation paragraph. The prosody control unit 50 performs speech synthesis by performing prosody control for each exhalation paragraph. In this way, data transmission for voice synthesis can be efficiently performed between communication data other than voice synthesis.
[0017]
Further, the input buffer 30A of the transceiver 30 in FIG. 1 may have a capacity for temporarily storing data transmitted for each exhalation paragraph. The prosody control unit 50 performs prosody control for each exhalation paragraph. For this reason, the capacity of the input buffer can be minimized. Note that the synthesized speech can be generated without interruption by starting the speech synthesis process when transmission of all data for each exhalation paragraph is completed.
[0018]
The transmission data from the information center side is controlled as “control command + prosodic pattern + word 1 (word reading + accent) + word 2 (word reading + accent) +... + Word n (word reading + accent)” A command may be provided. On the vehicle-mounted terminal side, prosody control and speech synthesis are performed according to this control command. With this control command, an option to transmit in units of one sentence (see FIG. 2) and an option to transmit in units of expiratory paragraphs (see FIG. 3) can be arbitrarily selected. Note that in the option of transmitting in the exhalation paragraph unit, the control command may be added only to the voice data in the first exhalation paragraph unit of one sentence.
[0019]
Furthermore, a voice type code such as male sound or female sound may be included in the control command and transmitted together with the prosodic parameters corresponding thereto. The prosody control unit 50 performs prosody control according to this prosody parameter, and the phoneme segment dictionary unit 60 selects and uses a phoneme dictionary that matches the gender's speech type, and uses the intended speech type on the information center side. Speech synthesis is possible.
[0020]
Moreover, you may enable it to request | require the audio | voice type of a male sound and a female sound from the vehicle-mounted terminal to the information center side. As a result, the information center transmits data together with the requested voice type code and the prosodic parameters corresponding thereto. On the in-vehicle terminal side, it is possible to synthesize speech with the required speech type by the same processing as described above.
In addition, about the above example, it can apply similarly not only for a weather forecast but the voice synthesis | combination of facility information, such as traffic information and a parking lot.
[0021]
【The invention's effect】
From the above description, according to the present invention, it is not necessary to add and improve the text to be inserted by the speech synthesizer on the in-vehicle terminal side and can be performed collectively on the information center side, so that the text can be easily added and improved. It becomes possible.
[Brief description of the drawings]
FIG. 1 is a diagram illustrating an example of an on-vehicle speech synthesizer according to the present invention.
FIG. 2 is a diagram illustrating an example of transmission data from the information center side.
FIG. 3 is a diagram illustrating another example of transmission data from the information center side.
FIG. 4 is a diagram showing a conventional on-vehicle speech synthesizer.
5 is a diagram illustrating an example of a prosodic pattern stored in a prosodic pattern storage unit 4 in FIG. 4;
[Explanation of symbols]
7 ... voice generation processing unit 10 ... inset synthesized data generation unit 20, 30 ... transceiver 30A ... input buffer 50 ... prosody control unit 60 ... phoneme segment dictionary unit

Claims (6)

In the in-vehicle speech synthesizer that synthesizes character string data transmitted from the information center side to the in-vehicle terminal side into speech,
On the information center side,
With accents plurality of word reading and said word constituting the character string data, and inset composite data generator for generating a temporal change information der Ru prosody pattern of frequencies in a breath,
A transmitter that transmits the character string data including the word reading and accent and the prosodic pattern data including the prosody pattern generated by the inset synthesized data generator ; and
On the in-vehicle terminal side,
A receiving unit for receiving the character string data and the prosodic pattern data transmitted from the information center side ;
The received by the reception unit, a vehicle-mounted voice synthesizer for the prosody control unit for prosody control by fitting the reading and the previous Kia accent before Kitan word to the prosody pattern, characterized in that is provided. The character string data is divided into exhalation paragraphs for voice synthesis ,
The inset composite data generation unit adds a control command to the transmission data to control whether the transmission data is transmitted in units of sentences or in units of expiratory paragraphs,
The in-vehicle speech synthesizer according to claim 1, wherein the prosody control unit performs prosody control according to the received control command . The prosody control unit performs prosody control at the end of reception of the speech pattern for the data in which the data related to the prosody pattern received for each exhalation paragraph, the reading of the plurality of words, and the accent of the word are temporarily stored. The in-vehicle speech synthesizer according to claim 2, wherein processing is started. A voice type code and a prosodic parameter corresponding to the voice type code are further transmitted to the character string data transmitted from the information center side, and the in-vehicle terminal side corresponds to the voice type code and the prosodic parameter. The in-vehicle speech synthesizer according to claim 1, wherein speech to synthesize is synthesized. The in-vehicle speech synthesizer according to claim 1, wherein the in-vehicle terminal side requests the information center side to select a voice type. In the in-vehicle speech synthesizer that synthesizes character string data transmitted from the information center side to the in-vehicle terminal side into speech,
On the in-vehicle terminal side,
A receiver for receiving the information from the information center side,
Time-dependent changes in frequency in character string data representing the reading of a plurality of words and accents of the words included in the reception information received by the reception unit, and an exhalation paragraph transmitted along with the character string data a data separator for separating the prosodic pattern is information, and
The separated by the data separating unit, a reproducing unit for reproducing the prosody control signal corresponding to the reading of the prosodic pattern or we each character string data, is provided,
A vehicle-mounted speech synthesizer characterized in that the generated speech is synthesized by sequentially controlling the prosody for the character string data separated by the prosodic control signal.

Images