JPH05210482A - Method for managing sounding dictionary - Google Patents

Abstract

PURPOSE:To reduce data capacity by efficiently preparing a sounding dictionary having no redundancy and to change the existent sounding dictionary into a dictionary saving uselessness. CONSTITUTION:When newly registering a word on a sounding dictionary 16, an accent link rule 12 is applied to the word and when the correct reading and accent of the word can be obtained, the word is not registered (steps 33, 34 and 35). When the correct reading and accent can not be obtained, the word is registered (steps 36 and 37). In the case of erasing a word registered on the sounding dictionary 16, the word is erased from the sounding dictionary 16 when the reading and accent can be obtained by applying the accent link rule 12 to plural registered words, and the word is not erased when the reading and accent can not be obtained.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pronunciation dictionary management method for registering and deleting words in a pronunciation dictionary used for Japanese text-to-speech conversion.

[0002]

2. Description of the Related Art Generally, a Japanese text-to-speech conversion system takes a kanji-kana mixed sentence that we read and write everyday as input, converts it into speech, and synthesizes and outputs it. The configuration of this system will be described with reference to FIG.

The text analysis unit 21 refers to the pronunciation dictionary 22 to generate a phoneme and a prosodic symbol string from the input Kanji / Kana mixed sentence. Here, the phoneme and the prosody symbol string are those in which the reading, accent, intonation, etc. of the input sentence are described as a character string (intermediate language). The reading and accent of each word are registered in the pronunciation dictionary 22, and a phoneme and a prosodic symbol string are generated with reference to this dictionary 22. Japanese has high and low accents, and when a position descending from high to low is at the nth syllable counted from the beginning of the word, the word is said to have an n-type accent. n = 0
Is a flat plate type when there is no descending position. In the pronunciation dictionary 22, as shown in FIG. 3, numerical values indicating the accent type for each word (excluding auxiliary words such as particles and suffixes), notation and reading of each word are registered.

Further, in the text analysis section 21, the accent combination rule shown in FIG. 4 is made into a rule. This is for the following reason. In Japanese, when plural words are combined to form a compound word, the phenomenon of accent movement, occurrence and disappearance occurs. Therefore, this phenomenon is made into a rule as an accent combination rule, and the accent of a compound word or a phrase is determined by applying this rule when combining words.

In the synthesis parameter generation unit 23, synthesis parameters such as a phoneme (sound type), phoneme duration (sound length), and basic frequency pattern (sound pitch) are set based on the phoneme and the prosodic symbol string. Is generated. The voice unit is a basic unit of voice for connecting and creating a synthesized waveform, and various units are stored as the voice unit storage unit 24 in accordance with the type of sound.
It is accumulated in OM etc.

The voice synthesizer 25 produces a synthesized waveform based on the synthesis parameter and outputs a synthesized sound through the speaker 26.

[0007]

By the way, the text analysis unit 21 determines the accent of a compound word or a phrase when combining words based on the accent combining rule of FIG. 4, and this accent combining rule is an accent when combining words. It represents the general trend of change, with some exceptions. Therefore, a compound word whose accent cannot be realized even if this rule is applied is regarded as one word in the pronunciation dictionary 2.
It is necessary to register in 2. On the contrary, if this rule is applied, it is not necessary to register the compound word that can obtain the accent.

However, since the pronunciation dictionary 22 is not created in consideration of such accent combination rules, compound words that do not need to be registered are also registered together with compound words that need to be registered. Sometimes In addition, when the accent combination rule is improved with the version upgrade of the system, a compound word that has been required to be registered may not be required to be registered thereafter.

In such a case, the words that need to be registered or no longer need to be registered wastefully use a part of the dictionary capacity, and the limited dictionary capacity cannot be used efficiently. There is a point.

The present invention has been made in view of the above problems, and an object thereof is to efficiently use a pronunciation dictionary by newly selecting words to be registered in the pronunciation dictionary and selecting registered words as necessary. The purpose is to provide a pronunciation dictionary management method.

[0011]

[Means for Solving the Problems] To solve the above problems, a first invention is to apply an accent combining rule to a word when a word is newly registered in a pronunciation dictionary, so that the word is correctly read and accented. The feature is that the word is not registered when is obtained, and is registered when it is not obtained.

In the second invention, when deleting a word registered in the pronunciation dictionary, when the pronunciation and the accent are obtained by applying an accent combination rule to a plurality of registered words, the word is pronounced. It is characterized by deleting from the dictionary and not deleting when it cannot be obtained.

[0013]

By the above method, the words that can be processed by the accent combination rule are not registered, and the already registered words that can be processed by this rule are deleted, so that the words newly registered in the pronunciation dictionary and the registered words can be registered as needed. It is possible to use the pronunciation dictionary efficiently by selecting it.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings.

The pronunciation dictionary management system according to the present embodiment is
As shown in FIG. 5, CRT 11 and accent combination rule 1
2, keyboard 13, control unit 14, text analysis unit 15
And a pronunciation dictionary 16.

The control unit 14 performs data exchange with the text analysis unit 15, registration management and deletion management of the pronunciation dictionary 16, input / output processing by the CRT 11 and the keyboard 13. The control unit 14 stores the processing functions shown in FIGS. 1 and 7 for performing registration management and deletion management.

The text analysis unit 15 uses the pronunciation dictionary 16 and the accent combination rule 12 to control the control unit 14.
The reading and accent of the text (word) sent from the player is determined, and the analysis result is returned to the control unit 14. As shown in FIG. 3, the pronunciation dictionary 16 includes each word (including compound words).
The notation, reading, accent, etc. are registered. As shown in the example of FIG. 4, the accent combination rule 12 is a rule that summarizes how the accent changes when words are combined. The registered contents of the pronunciation dictionary 16 and the accent combination rule 12 are the same as the conventional contents. The keyboard 13 is for inputting words registered in the pronunciation dictionary 16 and operation instructions to the system.
The CRT 11 displays readings, accents, etc., which are the results of text analysis.

Next, the processing (registration management method of the pronunciation dictionary 16) for registering words (particularly compound words) in the pronunciation dictionary 16 in the pronunciation dictionary management system configured as described above will be described with reference to the flowchart of FIG. .. Here, registration of “Ura Nihon (Uranihon: 4)” and “Sugar sugar (Kakuzato: 3)” is exemplified.

In the control unit 14, when these notations are input from the keyboard 13, the text analysis unit 1 receives them.
5 (step 31). The text analysis unit 15 performs the process of FIG. 6 described later.

Next, it is judged whether or not the analysis result is received from the text analysis unit 15 (step 32), and the process waits until it is received. When the analysis result is received from the text analysis unit 15, the analysis result is displayed on the CRT 11 (step 3
3) Ask the operator to confirm reading and accent.
Then, the operator confirms the display of the CRT 11 and waits for the input by the keyboard 13 (step 3).
4). This input is for instructing whether or not the display content on the CRT 11 is correct, and is input by "Yes" or "No".

Then, it is judged whether or not the keyboard input is "Yes" (step 35). If the answer is "Yes", it means that the analysis result [Uranihon: 4 [Ura Japan]] is correct, and there is no need for registration, and this process ends.

On the other hand, if "No" is input, it is found that the analysis result [Kakusato: 4 [sugar sugar]] is incorrect, and the input of correct reading and accent (Kakusato: 3) is waited for (step 36). When the correct reading and accent are input, the reading and accent are registered in the pronunciation dictionary 16 together with the notation (step 37), and the process ends.

Here, the processing in the text analysis unit 15 will be described with reference to FIG.

In step 31, the control unit 14 gives the notation ["Ura Nihon (4)" and "sugar sugar".
3) ”] is sent, these are divided into a plurality of words [back / Japan, Kaku / sugar] by referring to the pronunciation dictionary 16.
That is, morphological analysis is performed (step 41). Then, the accent combination rule 12 (in this case, the noun in FIG. 4 [1] +
The noun rule is applied to determine pronunciation and accent as follows (step 42).

Ura: 2 + Nihon: 2 → Ura Nihon: 4 [Back Japan] Kaku: 1 + Sato: 2 → Kaku Sato: 4 [Sugar] Then, the analysis result is transmitted to the control unit 14 (step 43). As a result, the control unit 14 displays the analysis result in step 33, and the subsequent processing is advanced.

Next, a process (deletion management method) for deleting a word (particularly a compound word) from the pronunciation dictionary 16 will be described with reference to the flowchart of FIG. Note that the registered "freighter (Kamotsusen: 0)" will be deleted here.

The control unit 14 extracts one word (word) from the pronunciation dictionary 16 and sends the notation [cargo ship] to the text analysis unit 15 (step 51). The text analysis unit 15 performs the process of FIG. 8 described later.

Next, it is judged whether or not a signal is received from the text analysis section 15 (step 52), and the process waits until a signal is received. When the signal is received, it is determined whether the signal is the notification of the analysis result or the notification that the word cannot be divided (step 53). If the notification indicates that the word cannot be divided, the process is terminated without deleting the word. If it is a notification of the analysis result, the analysis result of the word is compared with the reading and accent of the words extracted from the pronunciation dictionary 16 (the words corresponding to the individual words divided by the morphological analysis), and whether these are the same or not Is determined (step 54). This is a word that cannot be processed by the accent combination rule 12 unless the pronunciation and the accent are the same, and this processing is terminated without deleting.

On the other hand, if the reading and the accent are the same,
It can be processed by the accent combination rule 12 and does not need to be registered in the pronunciation dictionary 16. The word "cargo ship (Kamotssen: 0)" is deleted from the pronunciation dictionary 16 and the processing ends ( Step 55).

Here, the processing in the text analysis section 15 will be described with reference to FIG.

When the notation [cargo ship] is sent from the control unit 14 in step 51, these are pronounced in the pronunciation dictionary 16
Refer to to divide into multiple words [cargo / ship]. That is,
Morphological analysis is performed (step 61). As a result of the morphological analysis, it is determined whether or not the word can be divided into a plurality of words (step 62), and when the word cannot be divided into a plurality of words, the fact is notified to the control unit 14 (step 63), and the process is ended. .. If the division is successful, the accent combination rule 12 (in this case, the rule of noun + suffix in FIG. 4 [2]) is applied to determine the pronunciation and accent as follows (step 6).
4).

Kamotsu: 1 + Sen → Kamotsu Sen: 0 [cargo ship] Then, the analysis result is transmitted to the control unit 14 (step 65). As a result, the control unit 14 determines whether or not the notification of non-division is possible in step 53, and the subsequent processing is advanced.

As described above, by registering necessary words and deleting unnecessary words, the pronunciation dictionary 16 having no redundancy in terms of reading and accent can be efficiently created. That is, the data capacity of the pronunciation dictionary 16 can be reduced.

Further, the ready-made pronunciation dictionary 16 can be recreated into a lean dictionary.

When the Japanese text-to-speech conversion system is commercialized, the pronunciation dictionary 16 is often stored in ROM, in which case it cannot be rewritten. Therefore,
The pronunciation dictionary management system described in the present embodiment can be used as a development tool before the pronunciation dictionary 16 is stored in ROM.

However, the user dictionary that can be rewritten by the Japanese text-to-speech conversion system (user-specific pronunciation dictionary)
In case of having the above, the pronunciation dictionary management system may be incorporated into the Japanese text-to-speech conversion system and used as a management tool for this dictionary. As a result, the pronunciation dictionary 16 without redundancy can be created, and the ready-made pronunciation dictionary 16 can be created.
When incorporated into, it can be recreated into a lean dictionary.

[0037]

By using the pronunciation dictionary management method of the present invention, it is possible to efficiently create a pronunciation dictionary having no redundancy in terms of reading and accent.

[Brief description of drawings]

FIG. 1 is a flowchart showing a processing procedure of dictionary registration according to a pronunciation dictionary management method of the present invention.

FIG. 2 is a block diagram showing a conventional Japanese text-to-speech conversion system.

FIG. 3 is an explanatory diagram showing the contents of a pronunciation dictionary.

FIG. 4 is an explanatory diagram showing accent combination rules.

FIG. 5 is a block diagram showing a pronunciation dictionary management system according to the pronunciation dictionary management method of the present invention.

FIG. 6 is a flowchart showing a process at the time of dictionary registration in the text analysis unit.

FIG. 7 is a flowchart showing a processing procedure for word deletion.

FIG. 8 is a flowchart showing a process when deleting a word in the text analysis unit.

[Explanation of symbols]

 11 CRT 12 Accent Coupling Rule 13 Keyboard 14 Control Section 15 Text Analysis Section 16 Pronunciation Dictionary

Claims (2)

[Claims] 1. When registering a new word in the pronunciation dictionary,
A pronunciation dictionary management method characterized by applying an accent combination rule to a word, not registering the word when the correct reading and accent of the word are obtained, and registering when the word cannot be obtained. 2. When deleting a word registered in the pronunciation dictionary, when the pronunciation and accent of the word are obtained by applying an accent combination rule to a plurality of registered words, the word is deleted from the pronunciation dictionary. A pronunciation dictionary management method characterized by not deleting when not in use.