JPH09330093A - Voice synthesizing device synthesizing received character data and voice - Google Patents

Abstract

PROBLEM TO BE SOLVED: To simplify processing by which character data constituting a received KANJI and KANA sentence is synthesized to a voice. SOLUTION: In a voice synthesizing device by which a KANJI and KANA sentence for display is received and synthesized to a voice, in a transmitting side 20, a KANJI and KANA sentence for display is divided into words, character symbol train data for display, voice symbol train data for phonetics, and pause data, for each word are sent out, in receiving side 10, data sent from the transmitting side is received, the voice symbol train data for each word is extracted, and a voice is synthesized by adding rhyme information to this data.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a voice synthesizing apparatus used for a receiver mounted on a vehicle, and in particular, it receives FM character multiplex broadcasting and character data constituting VICS (Vehicle Information Communication System) information as voice. The present invention relates to simplifying the process of combining.

[0002]

2. Description of the Related Art FIG. 5 shows a receiver mounted on a vehicle.
Note that the same components are denoted by the same reference numerals or symbols throughout the drawings. As shown in this figure, as a receiver 10 installed in a vehicle, for example, in an FM radio receiver 1 that receives FM character multiplex broadcasting, a broadcast signal is a speaker 2
Is converted into voice. On the other hand, the character data in which the broadcast signal is multiplexed is displayed on the display 3. Then, the character data is synthesized into a voice signal by the voice synthesizing device 4 and converted into voice through the speaker 5.

In place of the above FM character multiplex broadcasting, V
Similarly, when the ICS information is received, the character data is
It can be displayed by a display device and converted into voice by a voice synthesizer. FIG. 6 is a diagram showing the speech synthesizer 4 of FIG. As shown in the figure, the speech synthesizer 4 divides the input character data of a character / symbol string into words using a grammatical interpretation and a semantic analysis for a mixed kanji / kana sentence, and outputs the characters of each word by phonetic expression. In order to convert into, that is, in other words, the language processing unit 41 that forms phonological information in which the data of the character symbol string is converted into the data of the phonetic symbol string is provided. In the processing of the language processing unit 41, a language that stores a phonological rule that converts a divided word from a character symbol string into a phonetic symbol string (reading) and a prosodic rule that gives prosodic information such as accent, intonation, and pause to the word. The dictionary 42 is used. In this way, accents, intonations, pauses, and the like are added, of course, so that a prose-like kanji-kana mixed sentence suitable for display is suitable for prosodic phonetic. The language dictionary 42 is composed of a ROM, and usually has a memory capacity of about 1 Mbit. The acoustic processing unit 43, which receives the voice symbol string processed by the language processing unit 41, converts the voice symbol string into a voice unit of a voice synthesis unit. In the processing of the phoneme processing unit 43, a phoneme dictionary 44 which stores a phoneme unit and is composed of a ROM is used.

By connecting the speech units obtained through this conversion, a continuous speech waveform is output, and the input character data as shown below is read aloud. Since the language processing unit 41 has a large amount of calculation, it is generally a general-purpose CP.
Since U is used and the acoustic processing unit 43 is required to have an arithmetic processing speed, a DSP high-speed arithmetic processor is used.

Input character data of kanji and kana mixed sentences "Hanshin Expressway up traffic jam Ikutagawa 7 km" Voice symbol string data "Hanshin Kosoku, Novoli, Jutai, Ikugawa, Nanakm"

[0006]

By the way, in the above speech synthesizing apparatus, the received character / symbol string data of the kanji / kana mixed sentence is converted into the data of the speech symbol string by using the huge language dictionary 42. For this reason, each receiver has to have a huge language dictionary and search the language dictionary to obtain phonological information and prosody information, which is a heavy load.

[0007] Further, even in the FM text multiplex broadcasting, in traffic information and weather forecast, since the information is visually transmitted, it is often a prose sentence, which is unnatural even if it is read aloud as it is. There's a problem. Therefore, in view of the above problems, it is an object of the present invention to provide a voice synthesizing device for synthesizing received character data into voice, which can reduce the load on the receiver and can form a natural sentence. .

[0008]

SUMMARY OF THE INVENTION In order to solve the above problems, the present invention provides a voice synthesizer for receiving a kanji mixed kana sentence for display and synthesizing it into a voice. The kana mixed sentence is divided into words, and the character symbol string data for display, the phonetic symbol string data and the pause data for each word are transmitted, and the receiving side receives the data transmitted from the transmitting side, The received character data is characterized in that the phonetic symbol string data for each word is extracted, and prosodic information is added to the accent and intonation to synthesize it into a speech. By this means, on the broadcast side,
Since a phonetic symbol string for word phonetic is added, it is unnecessary for the receiving side to have a large-capacity language dictionary.
It is not necessary to perform language processing for dividing a kanji-kana mixed sentence into words, and the burden on the receiving side can be greatly reduced.

The transmitting side further transmits number data indicating the sentence pattern of the kanji / kana mixed sentence for display, and the receiving side corresponds to the sentence pattern number data, and a phonetic symbol string of particles, auxiliary verbs and nouns. The data pattern is stored, and the phonetic symbol string data of the word is fitted into this pattern. By this means, it is possible to add particles, auxiliary verbs, nouns, etc. between words to make the display of prose-like kanji and kana mixed sentences more suitable for prosodic phonetics.

The transmitting side further transmits order data for voice synthesis for each word constituting the kanji / kana mixed sentence for display, and the receiving side voice synthesizes the words according to the order data. . By this means
It is possible to rearrange the order of the words that compose a kanji-kana mixed sentence suitable for display so that the sentence can be further adapted to prosodic phonetics.

On the transmitting side, word accent data is further transmitted, and on the receiving side, intonation is added to the received accent data to synthesize a voice. By this means, it is not necessary for the receiving side to have an accent prosody rule, so the load on the receiving side can be further reduced. On the transmitting side, a flag is attached to the data to be transmitted so that the receiving side can discriminate. By this means, processing can be performed based on this flag, and the processing on the receiving side is further reduced.

[0012]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows an FM broadcast or VICS transmitted by an FM broadcast station or VICS center according to the present invention.
It is a figure explaining the content of information. As shown in this figure, F
In the M broadcasting station or VICS center 20, at the same time as the character / symbol string data of each word that constitutes a kanji / kana mixed sentence to be transmitted, the voice symbol string data of each word, the voice synthesis order number, and the auxiliary data read number are As such, these are transmitted to the onboard receiver 10 of the vehicle.

Sending data example FF 213 OA 1 OB Hanshin high speed OC Hanshin Shinsok OA 2 OB upstream OC Novoli OA 5 OB traffic jam OC Jutai OA 3 OB Gyodagawa OC Ikugawa OA 4 OB 7 km OC Nana Kilometer Explanation of sending data example FF: Auxiliary data call number start flag 213: Auxiliary data call number OA: Speech synthesis sequence start flag OB: Kanji / Kana mixed sentence data start flag of character / symbol string OC: Voice symbol string data start flag 1, 2, 3, 4, 5: Speech Synthesis Sequence Number FIG. 2 is a diagram showing a speech synthesis apparatus for synthesizing received character data into speech according to the present invention. As shown in this figure, F
Based on the input data transmitted from the M broadcasting station or VICS center 20, the control unit 45 refers to the auxiliary data calling number and searches the auxiliary data dictionary 42 for auxiliary data corresponding to the sentence pattern of the kanji / kana mixed sentence. The auxiliary data dictionary 42 stores a large number of auxiliary data including patterns of particles, auxiliary verbs and nouns, as described below, in order to fit phonetic symbol string data of words in a kanji / kana mixed sentence.

FIG. 3 shows a phonetic sound stored in the auxiliary data dictionary.
It is a figure which shows an example of the auxiliary data of. As shown in this figure
And, regarding auxiliary data call number 213,
In response to "Dourono "," Wa (ha) "," De "
"No , And auxiliary data consisting of the phonetic symbol sequence of "death"
Is stored. In this ancillary data, the first
Rono Is a phonetic symbolic string of roads (nouns)
"No" is a particle, "Indicates a pause between words. Next
Is followed by three particles and a pose, and finally consists of auxiliary verbs
You. In this way, a kanji-kana mixed sentence suitable for display is displayed.
Insert particles, auxiliary verbs, and nouns between words according to the sentence pattern
From the prose display to the prosodic phonetic.

FIG. 4 illustrates an example of a control operation for fitting the phonetic symbol string data of each word into the auxiliary data by the auxiliary data calling number. As shown in the figure, the control unit 45 calls the auxiliary data calling number 213 from the auxiliary data dictionary 46 to the outline of the input data, and fits the phonetic symbol string data of the word to be sent out according to the phonetic synthesis sequence number. In this way, the order of the words constituting the kanji-kana mixed sentence suitable for display is changed so as to be suitable for prosodic phonetic.

Furthermore, the auxiliary data dictionary 42 stores prosody rules such as word accents and intonations for such auxiliary data. For example, regarding word accents, there are generally 0 type, 1 type, ..., N type and (n + 1) types of basic accent types in the n-mora proper noun phrase. The rules for type are used. Here, a mora is a unit that represents the length of a word, and one mora is approximately 5 in Japanese.
It corresponds to one character of 0 sound. In addition, when speaking Japanese, this is used as a rule because the intonation is downward sloping. Then, the control unit 45 combines the accent and the intonation.

Therefore, according to the present invention, on the side of the broadcasting station or the VICS center 1, for each word constituting the kanji / kana mixture sentence to be transmitted, the character / symbol string data for display and the phonetic symbol for phonetic sound. Since there is no need to have a large linguistic dictionary on the receiver side, there is no need to have a language dictionary. In addition, language processing for dividing a kanji-kana mixed sentence into words is unnecessary. In addition, the words that make up the sent-out Kanji / Kana mixed sentence are in an order suitable for display, but since they are given a voice synthesis order, they can be arranged in an order suitable for phonetics, and prosodic display emphasizes prosody. Will be possible.

[Brief description of drawings]

FIG. 1 is a diagram for explaining the contents of FM broadcast and VICS information transmitted by an FM broadcast station or VICS center according to the present invention.

FIG. 2 is a diagram showing a voice synthesizing apparatus for synthesizing received character data into voice according to the present invention.

FIG. 3 is a view showing an example of phonetic auxiliary data stored in an auxiliary data dictionary.

FIG. 4 is a diagram illustrating an example of a control operation of fitting phonetic symbol string data of each word into auxiliary data according to an auxiliary data calling number.

FIG. 5 is a diagram showing a receiver mounted on a vehicle.

6 is a diagram showing the speech synthesizer 4 of FIG.

[Explanation of symbols]

 10 ... Receiver 20 ... FM broadcasting station, VICS center 42 ... Phoneme dictionary 43 ... Sound processing unit 45 ... Control unit 46 ... Auxiliary data dictionary

Claims (5)

[Claims] 1. A voice synthesizer for receiving a kanji / kana mixed sentence for display and synthesizing it into a voice, wherein the transmitting side divides the kanji / kana mixed sentence for display into words, and displays a character for display for each word. The symbol string data, the phonetic symbolic phonetic symbol string data and the pause data are transmitted, and the receiving side
A voice synthesizing device for synthesizing received character data into voice, characterized by receiving data sent from a transmitting side, extracting voice symbol string data for each word, adding prosodic information to the extracted data, and synthesizing into voice . 2. The transmitting side further sends out number data indicating the sentence pattern of the kanji-kana mixed sentence for display, and the receiving side corresponds to the sentence pattern number data, and a voice of a particle, an auxiliary verb, or a noun. The speech synthesizer for synthesizing received character data into speech according to claim 1, wherein a pattern of symbol string data is stored, and the speech symbol string data of the word is fitted into this pattern. 3. The transmitting side further transmits order data for voice synthesis for each word constituting the kanji / kana mixed sentence for display, and the receiving side voices the word according to the order data. The voice synthesizing apparatus for synthesizing the received character data according to claim 2, into voice. 4. The reception according to claim 1, wherein the transmitting side further transmits word accent data, and the receiving side synthesizes voice by adding prosody information to the received accent data. A voice synthesizer that synthesizes character data into voice. 5. The received character data according to claim 1, wherein the transmitting side attaches a flag to the data to be transmitted so that the receiving side can determine the data. A speech synthesizer that synthesizes speech into speech.