JP3219678B2 - Speech synthesizer - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a speech synthesizing apparatus, and more particularly to synthesizing a character string mainly composed of kanji sentences obtained by receiving FM multiplex text broadcasting or the like with high sound quality.

[0002]

2. Description of the Related Art The above-mentioned FM multiplex teletext broadcast received,
Traffic information mainly composed of kanji sentences is used for display. It is more convenient if the traffic information obtained from the receiver mounted on the car is output as a display and also as a sound. This is because the driver should not look away from the front as much as possible for safety.

As a device for synthesizing the traffic information with speech, there is a text-to-speech synthesizing device that can use, as an input text, a kanji-kana sentence of the traffic information.

[0004]

However, traffic information mainly obtained from kanji sentences obtained from FM multiplex teletext broadcasting is prepared for display, and the contents are clear at a glance. However, even if the speech is used as it is, particles, conjunctions, verbs, and the like are omitted, and there is a problem that it is difficult to hear as spoken words.

SUMMARY OF THE INVENTION Accordingly, an object of the present invention is to provide a voice synthesizing apparatus which makes it easy to hear traffic information mainly composed of Chinese characters obtained from FM multiplex teletext, in view of the above problems.

[0006]

SUMMARY OF THE INVENTION In order to solve the above-mentioned problems, the present invention provides a method for linguistically processing text and converting the text to speech symbols.
In a speech synthesizer that performs columnarization , adds prosody, and synthesizes speech, a word that stores kanji and symbols that mainly constitute traffic information obtained from FM multiplex teletext and attribute information indicating the type of each word is stored. A dictionary, a language processing unit for performing language processing using the traffic information as text and extracting a word string with attributes, a plurality of predetermined basic attribute patterns, and grammatically connecting the basic attributes of the basic attribute patterns. And an inset synthesizing unit for searching for a basic attribute pattern to which the attribute of the extracted word string belongs, and for inserting the word into the inlaid pattern based on the search result. By this means, the words of the kanji sentence that mainly constitute the traffic information are inserted into the inset pattern composed of particles, conjunctions, verbs, etc., and the sentence becomes a sentence that is easy to hear as voice information.

The inset synthesizing unit converts two words connected in the word string into one word by using the preceding word as an adjective of the following word. By this means, it is possible to fit in various expressions from FM multiplex teletext broadcasting. For the converted word, a series of phonetic symbols
And perform prosody. By this means, it is possible to give optimal prosody corresponding to various expressions. In the search by the inset synthesizing unit, even if the attribute of the extracted word string does not match the basic attribute pattern, if the closest one exists, it is assumed to belong to this. This measure makes it possible to broaden the scope of the fit.

In the search of the inset synthesizing unit, it is assumed that even if the attribute of the extracted word string does not match the basic attribute pattern, if there is a closest one, it belongs to this. This measure makes it possible to broaden the scope of the fit. In the search of the inset synthesizing unit, if the order of the extracted word string is different from the arrangement order of the basic attribute patterns, the attribute is regarded as belonging to the rearranged order. This measure likewise makes it possible to increase the coverage of the fit.

[0009]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a diagram illustrating a speech synthesis device according to the present invention. FIG. 1 shows an antenna 1 for receiving FM multiplex teletext, an FM receiver 2 for demodulating a received signal, a speaker 3 for outputting FM broadcast sound in the demodulated signal, and a kanji sentence of traffic information in the demodulated signal. , A voice synthesizer 5 that makes it easier to hear kanji sentences and the like of traffic information and synthesizes the voice, and a synthetic sound from the voice synthesizer 5 and a broadcast sound from the FM receiver. And a switch 6 for selecting

FIG. 2 is a diagram showing a schematic configuration of the speech synthesizing unit 5 of FIG. As shown in the figure, the speech synthesis unit 5
A language analysis unit 51 that inputs teletext data from a receiver to generate a kanji sentence or the like into a word string mixed with word attributes, and a word that includes attribute information indicating the type of word and provides this information to the language analysis unit 51 A dictionary 52, an inset synthesizing unit 53 for performing an intelligible inset synthesis of the word string generated by the language analysis unit 51 as described later, and a speech sequence formation unit 54 for converting the inset and synthesized sentences into speech symbols. And a prosody provision unit 55 for providing information on reading kana, accent information, pose, and intonation.

FIG. 3 is a view for explaining an example of attributes assigned to words in the word dictionary 52 of FIG. As shown in the figure, a word representing a unique place name is given an attribute A, such as Ihagi A, Ishikawa-cho A, Bay Bridge A, and Haneda A. An attribute B is given to a word representing a general point, for example, a railroad crossing B. An attribute C is given to a word representing a unique route name, for example, Yokoha C, Kanpachi C, or the like. An attribute D is given to a word representing a general road, such as a line D, for example. An attribute H is given to a word representing time, for example, 21 minutes H. An attribute I is given to the word representing the distance, for example, 1.5KI. Words indicating directions include, for example, upper F and inner F,
An attribute F is given (upward means “inward”, and inside means “inward”). A symbol representing → is given an attribute K, for example, → K. An attribute Z is given to the symbol representing (), for example, (Z,) Z.

The attribute assignment to the word dictionary 52 is generally performed so as to correspond to the types of words included in FM multiplex teletext. FIG. 4 is a flowchart for explaining the processing of the language analysis unit 51 and the inlay synthesizing unit 53 of FIG. 2, and FIG. 5 is a diagram showing a first example of the processing of FIG.
FIG. 5 is a diagram illustrating a second example of the processing in FIG. 4.

In step S1, a word attribute is assigned to an example of teletext data (see FIGS. 5 (a) and 6 (a)) input to the language analysis section 51 (see FIGS. 5 (b) and 6 (b)). Is performed. These teletext data examples are displayed on the display 4.
Will be displayed. In step S2, the teletext data is cut according to the K and Z attributes to form a new word string (FIG. 5).
(C), FIG. 6 (c)).

In step S3, "Iogi A railroad crossing B" is converted into the attribute of "Iogi railroad crossing B" as shown in FIG. 6D, within the range of the new word subjected to the cutting processing. This is because the point “railroad crossing B” may be described by the place name “Iogi A” . As shown in FIG. 5 (d) Similarly, in the case of "through L" is epithet in the place name "Bay Bridge A",
“Baybridge A via L” is converted to “Baybridge L”.

When the road "Line D" is described by the line name "Yokohama" as in "Yokohama C Line D", it is converted into "Yokohama Line D". In step S4, the word string with the new attribute converted is cut into words for each attribute (see FIGS. 5E and 6E). In step S5,
Teletext data is represented by a sentence example pattern of attributes (FIG. 5).
(D), see FIG. 6 (d)).

In step S5, a sentence example pattern is searched. FIG. 7 is a flowchart for explaining the sentence example pattern search process in step S5 in FIG. Step S
At 11, it is confirmed that the extraction of the sentence example pattern is completed. In step S12, using the basic attribute patterns up to nn, which are pre-stored in the inlay synthesizing unit 53, a search is made as to which basic attribute pattern the sentence example pattern belongs to. The sentence example pattern in FIG.
The sentence example pattern of FIG. 6 is closest to the pattern of sentence example 01, and matches the pattern of sentence example 02.

In step S13, words in the sentence example pattern are rearranged in accordance with the basic attribute pattern. In step S6, the following inset synthetic data is generated. In order to make the sentence easy to understand as the voice information, the inset synthesizing unit 53 stores the sentence examples 01, 02,.
, An inset pattern composed of particles, connections, verbs and the like is stored, and the rearranged words are inserted into this inset pattern as follows.

[0018] In the example of FIG. 5, the horizontal wings on rehearsal, via the Bay Bridge from Ishikawa-cho to <br/> Haneda it will take 21 minutes.
In the example of (underlined inset pattern) 6, the inner around the ring bees, 1.5K from Ihagi crossing
In m, outside around is the 1.5K m to Ihagi railway crossing.

The inset sentence thus obtained is output as speech. This makes it easier to hear,
The traffic information of the FM teletext is formed into a high-quality synthesized sound.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating a speech synthesizer according to the present invention.

FIG. 2 is a diagram showing a schematic configuration of a voice synthesizing unit 5 of FIG. 1;

FIG. 3 is a diagram illustrating an example of an attribute assigned to a word in the word dictionary 52 of FIG. 2;

FIG. 4 is a diagram showing a language analysis unit 51 and an inlay synthesis unit 53 shown in FIG.
6 is a flowchart for explaining the processing of FIG.

FIG. 5 is a diagram illustrating a first example of the processing in FIG. 4;

FIG. 6 is a diagram illustrating a second example of the processing in FIG. 4;

FIG. 7 is a flowchart illustrating a sentence example pattern search process in step S5 of FIG. 6;

[Explanation of symbols]

 5: Voice synthesis unit 52: Word dictionary 51: Language analysis unit 53: Inset synthesis unit

Continuation of the front page (56) References JP-A-4-82000 (JP, A) JP-A-4-270395 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G10L 13 / 00-13/08 H04B 1/16 G08G 1/09

Claims (4)

(57) [Claims] 1. A linguistic processing of a text and a series of phonetic symbols
A speech dictionary for storing kanji and symbol words that mainly constitute traffic information obtained from FM multiplex teletext and attribute information indicating the type of each word A language processing unit for performing language processing using the traffic information as text to extract a word string with attributes; a plurality of predetermined basic attribute patterns; and a grammatically connecting the basic attributes of the basic attribute patterns. Having a pattern, and searching for which basic attribute pattern the attribute of the extracted word string belongs to, including an inset synthesizing unit that insets the word in the inlaid pattern based on the search result, wherein the inset synthesizing unit includes: A speech synthesizer capable of converting two words connected in the word string into one word by using a preceding word as an adjective of a following word. 2. A phonetic symbol for the converted word.
Perform sequence into, and performing prosody grant, speech synthesis apparatus according to claim 1. 3. The search of the inset synthesizing unit, wherein even if an attribute of the extracted word string does not match the basic attribute pattern, if there is a closest one, it is regarded as belonging to this. The speech synthesizer according to claim 1. 4. In the search of the inset synthesizing unit, if the attribute of the extracted word string is different from the arrangement order of the basic attribute pattern, the extracted attribute is regarded as belonging to the same. The speech synthesizer according to claim 3, characterized in that: