JPH0297999A - Speech synthesizer - Google Patents

Abstract

PURPOSE:To obtain the effect of a rapid reading so that it can be easy to hear by referring to a syntax analysis result, extracting word and phrases to satisfy prescribed conditions and outputting the syntax analysis result of a configuration element related to the word and phrases to a speech synthesizing part. CONSTITUTION:The character string of a sentence is inputted to a syntax analyzing part I, a syntax analysis is executed, and the result is outputted to an analysis result control part 3. At such a time, when an ordinary synthesis command is inputted from a setting part 2, the syntax analysis results as to all character string inputted from the analyzing part I are outputted to a speech synthesizing part 5 as they are. When a rapid traverse synthesis command is inputted, the analysis results related to the word and phrases made into the subject, object and important modifier of each sentence are successively extracted from the result of the analyzing part I and outputted to the speech synthesizing part 5. The speech synthesizing part 5 generates the parameters of the pitch pattern, power, etc., of the subject, object and important modifier for each sentence at the time of the rapid traverse synthesis and synthesizes the speech. In such a way, the effect of the rapid reading can be obtained so that it can be easy to hear.

Description

DETAILED DESCRIPTION OF THE INVENTION <Industrial Application Field> The present invention relates to a speech synthesizer that generates rule-based synthesized speech.

<Prior art> Conventionally, in the rule-based synthesis of speech using a speech synthesizer, when an arbitrary input character string is read quickly by synthesized speech, the phoneme to be generated is determined for all input character strings. This is done by shortening the duration of the .

<Problems to be Solved by the Invention> However, when performing fast reading using rule synthesis in conventional speech synthesis devices, it is necessary to omit input character strings and extract only the words and phrases required by the operator. This is done by shortening the duration of the phoneme, so
The synthesized speech that is output has a fast tone and is less natural. Therefore, there is a problem in that it is difficult to hear the synthesized speech when performing fast reading, and it is difficult to understand the content. Therefore, even if you increase the speed of quick reading to shorten the time, you will not get a commensurate effect.

SUMMARY OF THE INVENTION Therefore, an object of the present invention is to provide a speech synthesis device that can perform fast reading that makes it easy to hear and understand the content correctly.

<Means for Solving the Problems> In order to achieve the above object, the present invention divides an input character string into morphemes using a syntactic analysis unit, performs syntactic analysis, and generates speech based on the obtained syntactic analysis results. In a speech synthesizer that generates synthesis parameters necessary for speech synthesis in a synthesis section, and synthesizes and outputs speech according to these synthesis parameters, a setting section for setting normal synthesis or fast-forward synthesis, and a syntax from the above-mentioned syntax analysis section are used. The analysis results are taken in, and according to the normal synthesis or fast-forward synthesis set by the setting section, in the case of normal synthesis, the syntactic analysis results of all morphemes are output as they are to the speech synthesis section, while in the case of fast-forward synthesis, In this case, the speech synthesis unit includes an analysis result control unit that refers to the syntactic analysis results to extract words that satisfy a predetermined condition, and outputs the morpheme syntactic analysis results regarding the extracted words to the speech synthesis unit. It is a feature.

<Operation> When a character string is input to the syntax analysis unit, the character string is divided into morphemes and syntactically analyzed by this syntax analysis unit, and the obtained syntax analysis result is input to the analysis result control unit. . Here, when normal synthesis or fast-forward synthesis is set by the setting section, if normal synthesis is set, the analysis result control section outputs the syntactic analysis results of all morphemes as they are to the speech synthesis section. On the other hand, when fast-forward synthesis is set, the analysis result control section refers to the syntax analysis results and extracts words that meet predetermined conditions.
The results of syntactic analysis of morphemes related to the extracted words are output to the speech synthesis section.

Then, the speech synthesis section generates synthesis parameters necessary for speech synthesis based on the syntactic analysis results of the input morphemes, and speech is synthesized and output according to the synthesis parameters.

Therefore, in the case of normal synthesis, a synthesized speech of a sentence using all input character strings is output, and in the case of fast-forward synthesis, a synthesized speech of a sentence only containing words satisfying the above-described predetermined conditions is output.

<Example> Hereinafter, the present invention will be explained in detail with reference to illustrated embodiments.

FIG. 1 is a block diagram of a speech synthesizer which is an embodiment of the present invention. This speech synthesis device extracts the subject, predicate, and important modifiers of each sentence from any input sentence, and generates synthesized speech based on the extracted subject, predicate, and important modifiers. be.

When a character string of an arbitrary sentence is input to the syntactic analysis unit 1, the syntactic analysis unit 1 performs syntactic analysis of the input sentence and outputs the result of the syntactic analysis to the analysis result control unit 3. Then, the analysis result control section 3 controls the syntax analysis result as follows and outputs it to the speech synthesis section 5 according to the command from the setting section 2.

That is, when a normal synthesis command is input from the setting section 2,
The syntax analysis results for all input character strings input from the syntax analysis section 1 are output as they are to the speech synthesis section 5. When a fast-forward synthesis command is input, the speech synthesis unit 5 sequentially extracts the syntactic analysis results regarding the subject, predicate, and important modifier words of each sentence from the syntactic analysis results from the syntactic analysis unit l. Output to.

Based on the syntax analysis result inputted from the analysis result control section 3, the speech synthesis section 5 performs a synthesis process for all input character strings in the case of a normal synthesis command, and for each sentence in the case of a fast-forward synthesis command. Parameters necessary for speech synthesis, such as pitch pattern and power, are generated only for words and phrases that are subjects, predicates, and important modifiers, and speech is synthesized and output based on these parameters.

Next, the operation of the synthesized speech apparatus having the above configuration will be explained using specific example sentences. In this case, input text examples are written in brackets.

First, a character string of a sentence as shown below is input to the syntax analysis unit l.

“Once upon a time, there lived an old man and an old woman.One day, the old man went to the mountain to cut brushwood.The old woman went to the river to wash clothes.The old woman was washing clothes in the river. Then, a large peach came floating out of the field.The old woman picked up the large peach and took it home.''Then, the parser l parses the input string like the one above. and send the results to the analysis result control unit 3.
Output to. This syntax analysis is performed, for example, as follows. First, an input character string is divided into morpheme strings such as word strings with reference to a dictionary or the like, and morpheme analysis is performed to obtain the three pronunciations and part of speech of this morpheme. FIG. 2(a) shows the result of morphological analysis of the beginning part of the input character string. Next, based on the morphological analysis results, syntactic analysis is performed to determine the dependency relationship between each morpheme according to the dictionary and grammar rules. FIG. 2(b) shows the dependencies between clauses obtained based on the morphological analysis results shown in FIG. 2(a).

Next, for example, a normal synthesis command is input from the setting section 2. Then, based on this command, the entire input character string is synthesized into speech without omission of words and output. That is, the analysis result control unit 3 outputs the above-mentioned syntactic analysis results for all input character strings as they are to the speech synthesis unit 5. Then, the speech synthesis unit 5 generates parameters necessary for synthesizing pitch patterns, power patterns, etc. based on the syntactic analysis results regarding the input subject, predicate, and important modifiers, and according to these parameters, the speech synthesizer 5 generates the following speech. Start composing and outputting.

``Once upon a time, there lived an old man and an old woman in a certain place.'' Next, when the voice has been synthesized and outputted, for example, ``They lived...'', the setting section 2 commands a fast-forward synthesis command. Suppose that is input.

Then, the analysis result control unit 3 extracts words that serve as subjects, predicates, and important modifiers for each sentence based on the above-mentioned syntactic analysis results in the syntactic analysis unit l, and performs syntactic analysis on the extracted words. The result is output to the speech synthesis section 5. That is, in the example of the syntactic analysis results shown in Figure 2, based on the morphological analysis results in Figure 2(a) and the dependencies between morphemes in Figure 2(b), ``Ojisanwa'' is the subject and ``Ojisanwa'' is the subject. ``I went to the mountain'' is extracted as a predicate, and ``to the mountain'' is extracted as an important modifier.

At this time, important modifiers are extracted according to, for example, the following rules. In other words, select important modifiers of 1g or less per sentence. The selection group q is the attached word "wo"
The modifiers with the attached word "ni" or "he" are given priority and extracted as important modifiers, and the modifiers with the attached word "ni" or the attached word "he" are then extracted with priority.

If there are multiple modifiers with the same priority, the modifier closest to the beginning of the sentence is extracted as the important modifier. Therefore, in the case of the above input sentence example, "Fyama ni" and "Shibagari ni" both have the adjunct "ni", so they have the same priority, and "Yama ni" near the beginning of the sentence is considered an important modifier. Extract it.

And the word extracted above is "Ojisan wa".

The syntactic analysis results regarding ``I went to the mountain'' and ``Yama ni'' are output to the speech synthesis unit 5.

Then, the speech synthesis unit 5 generates parameters necessary for synthesis such as a pitch pattern and a power pattern for each phoneme with a normal duration length according to the syntactic analysis results regarding the input subject predicate and important modifiers, Speech is synthesized according to this parameter. Furthermore, in the same way, the subject, predicate, and important modifiers are extracted for each sentence one after another, and the following speech is synthesized and output according to the syntactic analysis results regarding the extracted subject, predicate, and important modifiers. do.

r Grandpa went to the mountain. Grandma went to the river. When the old woman was washing clothes, peaches came flowing down.

The old woman picked up some peaches and brought them home. ” Therefore, words other than the subject, predicate, and important modifiers are omitted from the input text mentioned above, and quick reading is performed using only the necessary words.

In this way, in this embodiment, when fast-forward synthesis is commanded, the subject, predicate, and important modifiers are extracted for each sentence from the input character string, and the generated words are generated based on the extracted words. To synthesize speech without shortening the duration of a phoneme. Therefore, during fast-forward synthesis, sentences output by synthesized speech are output at normal speed, and
The sentences are made up of only important words and phrases, so they are easy to listen to and understand the content correctly.

In the above embodiment, the subject, predicate, and important modifier are set as conditions for words to be extracted. However, the present invention is not limited thereto; for example, a sentence including a predetermined character string or the beginning and end of each sentence may be set.

Furthermore, it is also possible to easily convert the endings of the words extracted as described above into appropriate endings as needed and synthesize them into speech.

<Effects of the Invention> As is clear from the above, the speech synthesis device of the present invention has the following effects:
Normal synthesis or fast-forward synthesis is set by the setting section, and when normal synthesis is set, the analysis result control section outputs the syntax analysis results of all morphemes from the syntax analysis section to the speech synthesis section to generate speech. On the other hand, when fast-forward synthesis is set, the analysis result control unit outputs the syntactic analysis results of morphemes related to words that meet predetermined conditions to the speech synthesis unit to synthesize speech, so it is easy for the operator to Fast-forward synthesis can be performed by omitting the 35 unimportant phrases and using synthesized speech at normal speed.

Therefore, when fast reading is performed using the speech synthesis device of the present invention, the content can be easily heard and understood correctly, and the effects of fast reading can be sufficiently obtained.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of an embodiment of the speech synthesis device of the present invention, FIG. 2(a) is a diagram showing an example of a morphological analysis result, and FIG. 2(b) is a diagram showing an example of dependencies between clauses. It is a diagram. 1...Syntax analysis section, 2...Setting section, 3...
-Analysis result control unit, 5...speech synthesis unit.

Claims (1)

[Claims] (1) The input character string is divided into morphemes by the syntactic analysis unit, and syntactic analysis is performed. Based on the obtained syntactic analysis results, the speech synthesis unit generates synthesis parameters necessary for speech synthesis. In a speech synthesizer that synthesizes and outputs speech according to According to synthesis, in the case of normal synthesis, the syntactic analysis results of all morphemes are output as they are to the speech synthesis section, while in the case of fast-forward synthesis, the syntactic analysis results are referred to and a predetermined condition is satisfied. A speech synthesis device comprising: an analysis result control unit that extracts words and phrases and outputs a syntactic analysis result of morphemes related to the extracted words to the speech synthesis unit.