JP2010145873A - Text replacement device, text voice synthesizer, text replacement method, and text replacement program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To replace a word included in an input text with a word fit to a vocabulary ability of a hearing person. <P>SOLUTION: This text replacement device or the like sets a vocabulary ability level, analyzes the input text to input the word, determines whether the degree of word difficulty set preliminarily in each word exceeds the set vocabulary ability level or not (step S23), replaces the word when the degree of word difficulty exceeds the vocabulary ability level (step S24), and outputs the word as it is, without replacing the word, when the degree of word difficulty does not exceed the vocabulary ability level. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a text replacement device, a text-to-speech synthesizer, a text replacement method, and a text replacement program that can output synthesized speech by replacing words that are easy for a listener to understand.

  In general, in text-to-speech synthesis that generates synthesized speech from input text, words written in the input text are directly converted into speech and output. However, the input text is often written in written language assuming that the written characters are read directly by the reader's eyes. For this reason, when relatively difficult vocabulary and expressions are used, it is difficult for a listener who is listening to the ears without directly looking at the characters if the voice is changed as it is. In view of this, a text-to-speech synthesis method has been developed in which text written in written language is converted into spoken language expressed in plain expression and then synthesized speech is generated.

Japanese Patent Laid-Open No. 3-035296 (pages 3-5, FIGS. 1-2) JP 2002-149180 A (pages 4-8, FIGS. 1-7)

  However, in the conventional method, regardless of the listener's vocabulary, the written words are uniformly replaced with spoken words, so if the listener's vocabulary is high, the meaning becomes difficult to understand by replacing the words. There was a case. For example, the written contents can be communicated in a concise manner, but they can be difficult to understand due to rounded expressions.

  Also, whether the replaced word is easy to understand depends on the vocabulary of the listener. Therefore, even if it is changed to a vocabulary and expression that are generally easy to understand according to the unambiguous criteria, there are cases where it is still not understood by the listener.

  The present invention has been made to solve the above problem, and provides a text replacement device, a text-to-speech synthesis device, a text replacement method, and a text replacement program that can replace a word with a word according to vocabulary. The purpose is to do.

In order to achieve the above object, a text replacement device according to a first aspect of the present invention provides:
A text replacement device that replaces a word acquired from input text with a predetermined word,
A user interface to set and input vocabulary,
A storage unit for storing a word, a word equivalent to the word, and the vocabulary
A word is acquired from the input text, the acquired word is searched in the storage unit, and it is determined according to the vocabulary whether to replace the searched word with a predetermined word equivalent to the word. Processing means to
It is provided with.

The storage unit
For each word, along with the word, store the word difficulty index that indicates the difficulty of the word,
The processing means includes
Determining whether the word difficulty of the searched word exceeds the vocabulary;
When the word difficulty level exceeds the vocabulary, it is desirable to replace a word acquired from the input text and searched in the storage unit with a predetermined word.

The storage unit
For each word, a word candidate equivalent to the word to be replaced with the word is stored for each level of vocabulary,
The processing means includes
The storage unit may search for a word acquired from the input text and search for a vocabulary level to determine a candidate word to be replaced.

The user interface is
Set and enter the generation,
The storage unit
For each word, along with the word, remember the common sense of each word generation,
The processing means includes
Determining whether the common sense of the searched word in the input generation is below a predetermined threshold;
When the common sense of the searched word is lower than the predetermined threshold, the word acquired from the input text and searched in the storage unit may be replaced with a predetermined word.

In order to achieve the above object, a text-to-speech synthesizer according to a second aspect of the present invention provides:
Synthesizing speech based on text with words replaced by the text replacement device;
It is characterized by that.

In order to achieve the above object, a text replacement method according to a third aspect of the present invention includes:
A text replacement method for replacing a word obtained from input text with a predetermined word,
Set and input vocabulary,
Storing a word, a word equivalent to the word, and the vocabulary
Obtaining a word from the input text;
Searching for the acquired word;
Determining whether to replace a searched word with a predetermined word equivalent to the word, according to the vocabulary; and
It is provided with.

In order to achieve the above object, a text replacement program according to the fourth aspect of the present invention provides:
A text replacement program for replacing a word acquired from input text with a predetermined word,
On the computer,
Set and input vocabulary,
Storing a word, a word equivalent to the word, and the vocabulary
Obtaining a word from the input text;
Searching for the acquired word;
Determining whether to replace a searched word with a predetermined word equivalent to the word, according to the vocabulary; and
Is executed.

  According to the present invention, it is possible to provide a text replacement device, a text-to-speech synthesizer, a text replacement method, and a text replacement program that can replace a word with a word according to vocabulary.

Embodiments of the present invention will be described below with reference to the drawings.
First, the physical configuration of the text replacement device according to the present embodiment will be described with reference to FIG.

Embodiment 1
A text replacement device 100 shown in FIG. 1 includes a user interface 101, a storage unit 102, and a CPU (Central Processing Unit) 103.

  The user interface 101 is composed of, for example, a keyboard and a monitor, and the user of this apparatus can directly input various kinds of information. This is connected to the internal bus 105 via the interface circuit 104.

  The storage unit 102 includes a hard disk 106 and a RAM (Random Access Memory) 107.

  The hard disk 106 is composed of a magnetic disk, has a large capacity, and can retain stored contents even after the power is turned off.

  The RAM 107 is composed of a semiconductor memory and is volatile. The stored contents are retained only while the power is turned on, but can be accessed at a higher speed than the hard disk 106 and used as a work area for temporarily reading and processing the stored contents of the hard disk 106 while the power is turned on.

  The CPU 103 is mutually connected to the storage unit 102 via the internal bus 105, sequentially reads out the program stored in the storage unit 102, processes the data according to the command, and writes the result in the storage unit 102.

  Further, the interface circuit 104 is connected to the network 300. As a result of transferring the data from the data processing device (not shown) connected to the network 300 to the storage unit 102 via the network 300 and performing text substitution, the interface circuit 104 is obtained. Transferred data may be transferred to the data processing device.

  Next, the physical configuration of the text-to-speech synthesizer according to this embodiment will be described with reference to FIG.

  A text-to-speech synthesizer 200 is configured by adding a speaker 201 and a sound board 202 to the text replacement device 100.

  The speaker 201 need only be capable of outputting synthesized speech, and does not need to have high performance. The sound board 202 is provided between the speaker 201 and the internal bus 105, and outputs a signal for driving the speaker 201 corresponding to the synthesized voice data output from the CPU 103.

Next, the operation of the text replacement device 100 according to the first embodiment will be described.
In the apparatus of FIG. 1, the CPU 103 first stores the text replacement dictionary in the storage unit 102, then stores the vocabulary in the storage unit 102, and stores the input text in the storage unit 102. Then, text replacement is performed on the storage unit 102.

(Storage of text replacement dictionary in storage unit 102)
When the user issues a predetermined instruction through the user interface 101, the CPU 103 inputs a text replacement dictionary from the outside via the network 300 and stores it in the storage unit 102. The text replacement dictionary stores a word and a word equivalent to the word. For example, the contents shown in FIG. 4 will be described in the text replacement described later.

(Storing vocabulary in the storage unit 102)
When the user inputs vocabulary through the user interface 101, the CPU 103 stores this in the storage unit 102. Here, the vocabulary is a parameter that indicates how difficult the vocabulary the user can understand the meaning and usage. It is assumed that this parameter has higher vocabulary as the value is higher.

(Storage of input text to storage unit 102)
By receiving transfer data from another data processing apparatus (not shown) via the network 300, the CPU 103 stores the received transfer data in the storage unit 102. Upon receiving this, the CPU 103 performs text replacement.

(Text replacement on the storage unit 102)
In the text replacement shown in FIG. 2 (step S1), the CPU 103 acquires a word from the input text, performs morphological analysis (step S3), searches the storage unit 102 for the acquired word, and selects the searched word. It is determined according to the vocabulary level whether or not to perform word replacement (step S4) for replacing with a predetermined word equivalent to the word.

  FIG. 3 shows details of the processing in step S3 of FIG. The CPU 103 sets a counter j and initializes j = 1 (step S14). Then, using the morphological analysis dictionary 3 stored in the storage unit 102, morphological analysis is performed on the input text by any known method (step S15). Next, j-th morpheme information morpheme (j) is extracted from the result of the morpheme analysis (step S16).

Subsequently, it is determined whether or not the analysis has been completed up to the end of the input text (step S17). If it is determined that it has not been completed (step S17; No), the counter j is incremented by 1 to extract the next morpheme information (step S18), and the process returns to step S16.
On the other hand, if it is determined in step S17 that the analysis has been completed (step S17; Yes), the counter j value is moved to the numerical value Nmorpheme (step S19), and the morphological analysis is terminated.

  As a result, the input text is converted into a morpheme information string morpheme (k) (1 ≦ k ≦ Nmorpheme). Nmorpheme is acquired as the number of morphemes extracted from the input text.

  Next, details of word replacement (step S4) for the morpheme information string morpheme (k) (1 ≦ k ≦ Nmorpheme) will be described with reference to FIG.

  The CPU 103 sets a counter k and initializes k = 1 (step S21). Then, the morpheme information morpheme (k) is detected from the text replacement dictionary 4 stored in the storage unit 102 (step S22).

  Subsequently, the CPU 103 compares the word difficulty level VocabularyDifficulty (k) corresponding to the morpheme information morpheme (k) with the vocabulary level VocabularyLevel (step S23). And the following formula (1)

Word difficulty VocabularyDifficulty (k)> VocabularyLevel (1)
When satisfy | filling (step S23; Yes), CPU103 substitutes the morpheme information Morpheme (k) which is the word information 43 after substitution for the morpheme information morpheme (k) which is the word information 41 before substitution shown in FIG. S24).

  When the expression (1) is not satisfied (step S23; No), the CPU 103 does not replace the morpheme information morpheme (k) that is the pre-substitution word information 41 with the morpheme information Morpheme (k) that is the post-substitution word information 43. .

Subsequently, it is determined whether or not the counter k has reached Nmorpheme and all the morpheme information extracted by the morpheme analysis has been replaced (step S25). When it is determined that the replacement has not been completed for all morpheme information (step S25; No), the counter k is incremented by 1 (step S26), the process returns to step S22, and the processes after step S22 are performed again.
On the other hand, when it is determined that the replacement has been completed for all the morpheme information (step S25; Yes), the word replacement process is terminated.

  By performing the above processing, for example, the text substitution has been performed on the morpheme information sequence morpheme (k) (1 ≦ k ≦ Nmorpheme) of the input text “distorted and reported fact” shown in FIG. become. Here, Morpheme (k) (1 ≦ k ≦ Nmorpheme) is assumed to be a morpheme information sequence subjected to text replacement.

Next, the operation of the text-to-speech synthesizer according to this embodiment will be described.
In the case of a text-to-speech synthesizer, in FIG. 2, after text substitution (step S1), speech synthesis (step S2) is performed.

  In the speech synthesis (step S2), the CPU 103 performs prosody setting for the morpheme information string (step S5). In other words, prosody information is added to each Morpheme (k) included in the morpheme information sequence Morpheme (k) (1 ≦ k ≦ Nmorpheme) subjected to text substitution by any known method to label the label sequence with prosodic information Label ( m) (1 ≦ m ≦ N Label) is generated.

  Then, the speech synthesis dictionary 5 is used to synthesize speech for the label string with prosodic information Label (m) (1 ≦ m ≦ N Label), and a synthesized speech SynthesizeSpeech is output (step S6).

  With the above processing, as shown in FIG. 7, for example, the following operation is performed for a listener whose vocabulary level is “1”. For this listener, the word difficulty level 42 of “distortion” and “report” exceeds the vocabulary level. Then, “distortion” is replaced with “distortion”, and “report” is replaced with “shrimp”. As a result, the input text “Factual fact is distorted and reported” shown in FIG. 6 is output as a synthesized speech “Make the word distorted”, and a speech that is easy to understand for a listener whose vocabulary level is “1”. can get.

  Also, as shown in FIG. 7, for example, for listeners with vocabulary level “2” to “5”, only the word difficulty level 42 of “distortion” exceeds the vocabulary level, and only “distortion”. Is replaced with “distorted”. As a result, the input text “Report the fact is distorted” shown in FIG. Therefore, written words such as “report” that are used as spoken words by people with a vocabulary level of “2” or higher do not need to be rephrased and are left as they are. As a result, a voice that is easy to understand for listeners with vocabulary level "2" to "5" can be obtained.

  Furthermore, as shown in FIG. 7, for a listener with a vocabulary level of “6”, the word difficulty level 42 of all words does not exceed the vocabulary level, and all words are output as they are. As a result, the input text “Report the fact is distorted” shown in FIG. 6 is read as it is, and is output as a synthesized speech “I am going to make a mistake. As a result, it is possible to obtain a voice that is easy to understand for a listener whose vocabulary level is “6” and does not need to be reworded.

  In this way, synthesized speech that is suitable for each vocabulary level listener is output.

<< Embodiment 2 >>
In the second embodiment, a text replacement dictionary 6 shown in FIG. 8 is used instead of the text replacement dictionary 4 in the first embodiment shown in FIG. 2, and the text processing (step S1 in FIG. 2) is different from the first embodiment. Replace text with.

The physical configuration of the text-to-speech synthesizer according to the second embodiment is the same as that shown in FIG.
The text replacement dictionary 6 stores pre-substitution word information 61 and post-substitution word information 62.

  Pre-substitution word information 61 is pre-substitution word information obtained by morphological analysis. In the illustrated example, input text such as “facts”, “distortion”, “report”, etc. It is a word that is included in "Report."

  The post-substitution word information 62 is a post-substitution word corresponding to the pre-substitution word information 41. A plurality or a single candidate of this word is determined according to the vocabulary level of the listener who receives the synthesized speech.

  Next, the operation of the apparatus according to the second embodiment will be described with reference to FIGS. Note that FIG. 1 is referred to for the configuration of the apparatus of the present embodiment.

  First, the text replacement device 100 performs morphological analysis on the input text stored in the storage unit 102 (step S3 in FIG. 2). This is performed in the same procedure as the procedure according to the first embodiment shown in FIG.

  Next, the text-to-speech synthesizer 200 performs text substitution on the morpheme information string morpheme (k) (1 ≦ k ≦ Nmorpheme) according to FIG. First, the CPU 103 sets a counter k and initializes k = 1 (step S31). Then, the morpheme information morpheme (k) is detected from the text replacement dictionary 6 stored in the storage unit 102 (step S32).

  Subsequently, the CPU 103 replaces the morpheme information morpheme (k) with candidate morpheme information Morpheme (k) corresponding to the vocabulary level VocabularyLevel (step S33).

  Subsequently, it is determined whether or not the counter k has reached Nmorpheme (step S34). If it is determined that it has not been reached (step S34; No), the counter k is incremented by 1 (step S35), then the process returns to step S32, and if it is determined that it has been reached (step S34; Yes), the process ends.

  As a result, as shown in FIG. 7, for a listener whose vocabulary level is “1”, for example, “distortion” is replaced with “distorted” and “report” is replaced with “she let”. As a result, the input text “Factual fact is distorted and reported” shown in FIG. 6 is output as a synthesized speech that “distorts the truth” and is easy to understand for a listener whose vocabulary level is “1”. Is obtained.

  Also, as shown in FIG. 7, for example, only “distortion” is replaced with “distorted” for listeners with vocabulary level “2” to “5”. As a result, the input text “Report the fact is distorted” shown in FIG. Therefore, a written word that does not need to be reworded, such as “report”, which is used as a spoken word by a person whose vocabulary level is “2” or higher, is left as it is. As a result, a voice that is easy to understand for listeners with vocabulary level "2" to "5" can be obtained.

  Further, as shown in FIG. 7, all the words are output as they are to the listener whose vocabulary level is “6”. As a result, the input text “Report the fact is distorted and reported” shown in FIG. 5 is read as it is and is output with the synthesized speech “I am going to make a mess.” Voices that are easy to understand for listeners with a vocabulary level of "6" without speech.

  In this way, synthesized speech that is suitable for each vocabulary level listener is output.

<< Embodiment 3 >>
In the third embodiment, instead of the word difficulty level 42 in the text replacement dictionary 4 in the first embodiment shown in FIG. 4, the concept of common sense with respect to words for each generation is used, and text is processed in a processing procedure very similar to that of the first embodiment. Replacement (step S1 in FIG. 2) is performed. Here, common sense refers to the rate at which the vocabulary and expression of words, meanings, and usage are known within the same generation.

  That is, in the first embodiment, the word replacement is performed by indicating the difficulty level of the word at a rate known among all people regardless of the generation. On the other hand, in this embodiment, since the number of people who know the word among the people of that generation for each generation sharing the spoken word is different, it is necessary to know each word for each generation. Find the percentage of people who are. The ratio is adopted as an index of common sense in the generation.

  The text replacement dictionary 7 shown in FIG. 10 stores pre-substitution word information 71, common sense 72 for each generation, and post-substitution word information 73.

  Pre-substitution word information 71 is pre-substitution word information obtained by morphological analysis. In the example illustrated in FIG. 10, “facts”, “distortion”, “report”, and the like are words included in the input text, for example, “report the fact by distortion”.

  The common sense 72 for each generation is provided for each word stored in the text replacement dictionary 7 as the pre-substitution word information 71. The common sense 72 for each generation indicates, for example, a percentage value in which the meaning and usage of the word are known in the same generation.

  The post-substitution word information 73 is a word having a high level of common sense for each generation corresponding to the pre-substitution word information 71. In the example shown in the figure, the word “distorted” is replaced with “distorted” in any generation. In addition, for the word “report”, the words “replace” in the generations “A” and “B” are the replacement words, and the word “report” is the replacement word in the generation “C”. In addition, for the word “facts”, “jijitsu” is used as the replacement word in all generations, and the word is not replaced.

  Next, the operation of the text replacement device according to the third embodiment will be described with reference to FIG. Note that FIG. 1 is referred to for the configuration of the apparatus of the present embodiment.

  First, the CPU 103 sets a counter k and initializes k = 1 (step S41). Then, the morpheme information morpheme (k) is detected in the text replacement dictionary 7 stored in the storage unit 102 (step S42).

  Subsequently, the CPU 103 compares the common sense CommonSenseRate [Generation] (k) for each generation corresponding to the morpheme information morpheme (k) with a predetermined threshold value THCommonSense (step S43). And the following formula (2)

Common sense rate for each generation CommonSenseRate [Generation] (k) <predetermined threshold TH CommonSense (2)
It is determined whether or not the above is satisfied. Thereby, it is determined whether the common sense level is lower than a predetermined degree. For example, in the example of FIG. 10, if THCommonSense = 70%, “distortion” is determined to be low in common sense in generations A and B, and is determined as high in common sense in generation C. In addition, “report” is determined to be low in common sense in generation A, and is determined to be high in common sense in generations B and C.

  When the expression (2) is satisfied (step S43; Yes), the CPU 103 replaces the morpheme information morpheme (k), which is the pre-substitution word information 71 shown in FIG. 10, with the morpheme information Morpheme (k), the post-substitution word information 73. (Step S44).

  When the expression (2) is not satisfied (step S43; No), the CPU 103 does not replace the morpheme information morpheme (k) that is the pre-substitution word information 71 with the morpheme information Morpheme (k) that is the post-substitution word information 73. .

  Subsequently, it is determined whether or not the counter k has reached Nmorpheme (step S45). If it is determined that it has not been reached (step S45; No), the counter k is incremented by 1 (step S46), then the process returns to step S42, and if it is determined that it has been reached (step S45; Yes), the process ends.

  In the text replacement dictionary 7, for example, regarding the pre-substitution word information 71 of “distortion”, the common sense 72 for each generation is set to 10% for generation A, 50% for generation B, and 80% for generation C. The pre-substitution word information 71 “report” is set to 20% for generation A, 80% for generation B, and 80% for generation C. The pre-substitution word information 71 “facts” is set to 90% for generations A, B, and C.

  Accordingly, as shown in FIG. 12, for example, for a generation A listener, the common sense 72 of “distortion” and “report” falls below a predetermined threshold of 70%, and “distortion” becomes “distorted”. , "Report" is replaced with "she let". As a result, the input text “Report the fact is distorted and reported” shown in FIG. 6 is output as a synthesized speech “Make the word distorted”, and a speech easy to understand for the generation A listener is obtained.

  As shown in FIG. 12, for example, for the listener of generation B, the commonness 72 of “distortion” falls below a predetermined threshold of 70%, and only “distortion” is replaced with “distortion”. As a result, the input text “Report the fact is distorted” shown in FIG. Therefore, the written word “report” that is also used as a spoken word among generations B does not need to be reworded and is left as it is. As a result, a voice that is easy for a generation B listener to understand is obtained.

  Furthermore, as shown in FIG. 12, for the listener of generation C, the common sense level 72 of all words does not fall below a predetermined threshold value, and all the words are output as they are. As a result, the input text “Report the fact is distorted and reported” shown in FIG. 6 is read as it is and is output with a synthesized speech of “I am going to make a mistake.” It is necessary to rephrase the word. Sounds that are easy to understand for generation C listeners with no sound.

  In this way, synthesized speech adapted to each generation A, B, C listener is output.

  While the embodiments of the present invention have been described above, various modifications and applications can be made in implementing the present invention, and the present invention is not limited to the above embodiments.

  For example, in the above embodiment, morphological analysis and text replacement are performed separately, but they may be performed simultaneously. This is because the word difficulty level 42 corresponding to the pre-substitution word information 41, 61, 71 stored in the text substitution dictionaries 4, 6, 7 and 7, the post-substitution word information 43, 62, 73, and the common sense 72 for each generation, respectively. Is stored in the morpheme analysis dictionary 3 as information corresponding to the registered language stored in the morpheme analysis dictionary 3.

  In each of the above embodiments, the case where the apparatus of the present invention is configured by installing a program from a recording medium such as a CD-ROM into a general-purpose computer has been described. However, a ROM (Read Only Memory) that stores the program in advance is used. The device of the present invention may be configured by connecting to an internal bus.

It is a figure which shows an example of a structure of the text substitution apparatus and text-to-speech synthesizer of this invention. It is a figure which shows an example of the program structure performed with CPU in the apparatus of FIG. It is a flowchart which shows the procedure of the morphological analysis in FIG. It is a figure which shows an example of a structure of the text replacement dictionary in FIG. 6 is a flowchart illustrating a text replacement processing procedure according to the first embodiment. It is a figure which shows an example of an input text. It is a figure which shows an example of a synthetic | combination voice. It is a figure which shows the other example of a structure of the text replacement dictionary in FIG. 10 is a flowchart illustrating a text replacement processing procedure according to the second embodiment. It is a figure which shows the further another example of a structure of the text replacement dictionary in FIG. 14 is a flowchart illustrating a text replacement processing procedure according to the third embodiment. It is a figure which shows the other example of a synthetic voice.

Explanation of symbols

3 ... Morphological analysis dictionary, 4, 6, 7 ... Text substitution dictionary, 5 ... Speech synthesis dictionary, 41, 61, 71 ... Word information before substitution, 42 ... Word difficulty, 43 , 62, 73 ... Word information after substitution, 72 ... Common sense, 100 ... Text substitution device, 200 ... Text-to-speech synthesizer, 101 ... User interface, 102 ... Storage unit, DESCRIPTION OF SYMBOLS 103 ... CPU, 104 ... Interface circuit, 105 ... Internal bus, 106 ... Hard disk, 107 ... RAM, 200 ... Text-to-speech synthesizer, 201 ... Speaker, 202 ...・ Sound board, 300 ... Network

Claims (7)

A text replacement device that replaces a word acquired from input text with a predetermined word,
A user interface to set and input vocabulary,
A storage unit for storing a word, a word equivalent to the word, and the vocabulary
A word is acquired from the input text, the acquired word is searched in the storage unit, and it is determined according to the vocabulary whether to replace the searched word with a predetermined word equivalent to the word. And a processing means for performing the text replacement. The storage unit
For each word, along with the word, store the word difficulty index that indicates the difficulty of the word,
The processing means includes
Determining whether the word difficulty of the searched word exceeds the vocabulary;
When the word difficulty level exceeds the vocabulary, the word retrieved from the input text and searched in the storage unit is replaced with a predetermined word.
The text replacement device according to claim 1, wherein: The storage unit
For each word, a word candidate equivalent to the word to be replaced with the word is stored for each level of vocabulary,
The processing means includes
In the storage unit, searching for a word acquired from the input text, searching for a level of vocabulary, and determining a candidate word to replace,
The text replacement device according to claim 1, wherein: The user interface is
Set and enter the generation,
The storage unit
For each word, along with the word, remember the common sense of each word generation,
The processing means includes
Determining whether the common sense of the searched word in the input generation is below a predetermined threshold;
When the common sense of the searched word is lower than the predetermined threshold, replace the word acquired from the input text and searched in the storage unit with a predetermined word,
The text replacement device according to claim 1, wherein: Synthesizing speech based on text in which words are replaced by the text replacement device according to any one of claims 1 to 4;
A text-to-speech synthesizer characterized by the above. A text replacement method for replacing a word obtained from input text with a predetermined word,
Set and input vocabulary,
Storing a word, a word equivalent to the word, and the vocabulary
Obtaining a word from the input text;
Searching for the acquired word;
Determining whether or not to replace the searched word with a predetermined word equivalent to the word, according to the vocabulary ability. A text replacement program for replacing a word acquired from input text with a predetermined word,
On the computer,
Set and input vocabulary,
Storing a word, a word equivalent to the word, and the vocabulary
Obtaining a word from the input text;
Searching for the acquired word;
And determining whether to replace the searched word with a predetermined word equivalent to the word according to the vocabulary.

Images