JPS63187299A - Word spelling-pronunciation sign conversion system - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed Description of the Invention] [Summary] In the word spelling-one-phonetic symbol conversion method, each time one affix is separated from the spelling of a word or word stem according to a predetermined affix separation rule, the word stem part from which the affix has been separated becomes an exception. Search for exception words with similar pronunciations. Thereby, it is possible to speed up the word spelling-one-phonetic-symbol conversion process, and to prevent error-free word-spelling-one-phonetic symbol conversion by preventing unnecessary affix separation.

[Industrial application field]

The present invention relates to a word spelling-to-phonetic symbol conversion method that converts word spellings into phonetic symbols at high speed and without errors in a sentence-to-speech conversion process for converting arbitrary sentences into speech.

When converting an arbitrary sentence into speech, first, each word in the sentence is converted into a symbol representing its pronunciation, that is, a phonetic symbol.

However, there are many words that have exceptional pronunciations that do not follow regular conversion rules. For example, the spelling r
The regular pronunciation of oVEl is (o u v) or [
ΔV], but in exceptional cases it is pronounced as [U:v]. Examples of the latter include move (mu:v) and pro
There is ve (pru:v).

If a dictionary were to contain all the derivative forms of words with such exceptional pronunciation, it would be impractical because it would require a dictionary with an enormous storage capacity. Also, when new words are created, it is not possible to respond to them. Therefore, by having only the basic form (word stem) of the word in the dictionary for exception words, and searching the back foreign word dictionary that separates the word spelling into affixes such as suffixes and prefixes and the word stem, The storage capacity of the exception word dictionary is reduced, and a wide variety of exception words can be searched.

Furthermore, the pronunciation of the same word may change depending on the position of the accent. For example, the English spelling rxJ is pronounced as [g2] when the last vowel is accented.
In other cases, it is pronounced (ks). Therefore, when converting a word spelling into a phonetic symbol, it is necessary to determine the accent position of the word spelling and convert the word spelling into a phonetic symbol according to a predetermined pronunciation rule.

Therefore, when converting a word spelling into a phonetic symbol, the conversion process based on regular conversion rules and exceptional word processing are combined to efficiently perform the word spelling one phonetic symbol conversion process.

[Conventional technology]

FIG. 4 is a block diagram showing the configuration of a conventional word spelling and phonetic symbol conversion method.

In FIG. 4, reference numeral 30 denotes an exceptional word processing means, which processes words with exceptional pronunciation. In this exceptional word processing means 30, 31 is an exceptional word dictionary, which records the spelling of a word with an exceptional pronunciation and its phonetic symbol, or the spelling of a word to which an exceptional accent is added and the spelling with an accent added to that spelling. are stored in pairs.

A suffix separation rule dictionary 32 stores suffix separation rules for separating suffixes from word stems.

A prefix separation rule dictionary 33 stores prefix separation rules for separating prefixes from word stems.

Reference numeral 34 denotes a first exception word dictionary search unit, which performs processing to search whether the input word spelling exists in the exception word dictionary 31.

35 is a suffix separation processing unit, which includes a suffix separation rule dictionary 32;
With reference to , each suffix separation rule is applied to separate each suffix added to the word stem from the word stem.

Reference numeral 36 denotes a second exceptional word dictionary search unit, which performs processing to search whether a word stem from which a suffix has been separated exists in the exceptional word dictionary 31.

37 is a prefix separation processing unit, which includes a prefix separation rule dictionary 33;
, and performs processing to separate each prefix added to the word stem from the word stem by applying the separation rules for each prefix.

Reference numeral 38 denotes a third exceptional word dictionary search unit, which performs processing to search whether a word stem from which a prefix has been separated exists in the exceptional word dictionary 31.

Reference numeral 41 is an accent position estimation rule dictionary, which stores rules for adding an accent to the spelling of words.

Reference numeral 42 denotes an accent position estimating unit that refers to the accent position estimation rule dictionary 41 and performs a process of estimating the accent position for the input word spelling or the spelling that is a combination of an unconverted affix spelling and a phonetic symbol.

A pronunciation rule dictionary 43 stores rules for converting accented spellings into phonetic symbols.

44 is a spelling and phonetic symbol converter, and a pronunciation rule dictionary 43
With reference to , processing is performed to convert the input accented spelling or a combination of spelling and phonetic symbols into a character string with an accent added to the phonetic symbols.

Next, the word-phonetic symbol processing operation shown in FIG. 4 will be explained.

When word spelling data read from a sentence is input to the exception word processing means 30, the first exception word dictionary search unit 34 searches the exception word dictionary 31 to see if the spelling of the input word exists. If the word exists in the exception word dictionary 431 and is paired with a phonetic symbol, the phonetic symbol is output and the exceptional word processing ends. If the spelling is paired with an accented spelling, the accented spelling is input to the spelling-one-phonetic-symbol converter 44.

If the word does not exist in the exception word dictionary 31, that is, if the search fails, the spelling of the word is input to the suffix separation processing unit 35.

The suffix separation processing unit 35 refers to the suffix separation rule dictionary 32, applies separation rules for each suffix, and performs processing to separate each suffix added to the word stem from the word stem. When multiple suffixes are added to a word stem, all the added suffixes are separated by sequentially applying the separation rules for each suffix.

The second exception word dictionary search unit 36 performs processing to search whether the word stem from which the suffix has been separated exists in the exception word dictionary 31.

If the word exists in the exception word dictionary 31, the phonetic symbol of the word stem is read out and output as a character string combined with the separated suffix.

This character string output is composed of phonetic symbols (stem part) and unconverted spelling (suffix part), and since the accent position has not been determined, it is sent to the accent position estimating unit 42 to determine the accent position.

If the word does not exist in the exception word dictionary 31, that is, if the search fails, a character string consisting of a word stem and a suffix is input to the prefix separation processing unit 37.

The prefix separation processing unit 37 refers to the prefix separation rule dictionary 33, applies separation rules for each prefix, and performs processing to separate each prefix added to the word stem from the word stem. If multiple prefixes are added to a word stem, all the added prefixes are separated by sequentially applying the separation rules for each prefix.

The third exception word dictionary search unit 38 performs processing to search whether the word stem from which the prefix has been separated exists in the exception word dictionary 31.

If it exists in the exceptional word dictionary 31, the phonetic symbol of the stem is read out and output as a character string combined with the separated prefix and suffix. This character string output consists of phonetic symbols (stem part) and unconverted spellings (prefix and suffix parts), and the accent position has not been determined. Therefore, in order to determine the accent position, the accent position estimation unit 42 Sent.

If the word does not exist in the exception word dictionary 31, that is, if the search fails, this is a case in which the input word spelling is a word spelling that is converted into phonetic symbols according to regular conversion rules and is not an exception word.

In this case as well, the information is input to the accent position estimation unit 42 in order to determine the accent position of the unit spelling.

The accent position estimating unit 42 uses the character strings in which the affix spelling and phonetic symbol are combined, which are input from the second and third exceptional word dictionary searching units 36 and 38, and the third exceptional word dictionary searching unit 24.
For the word spellings input from 5, the accent position is estimated by referring to the accent position estimation rule dictionary 41, and the processed words are sent to the spelling and phonetic symbol conversion section 44.

The spelling/phonetic symbol conversion unit 44 refers to the pronunciation rule dictionary 43, and performs a process of converting the input accented spelling or the character string in which the spelling and the phonetic symbol are combined and accent information is added into the phonetic symbol.

In the manner described above, the spelling of a word can be converted into a phonetic symbol, including cases where the input word spelling is an exception word and cases where the spelling of the input word is converted into a phonetic symbol according to regular conversion rules.

[Problem that the invention seeks to solve]

Generally, the number of affixes connected to one word is not large. For example, in the case of English words, there are 0 to 3 prefixes and 0 to 4 suffixes. Therefore, among the many affix separation rules, the only one that can be applied to that word is Fleur.

However, since it is not known which rules can be applied, it is necessary to check the applicability of all rules in the suffix and prefix separation rule dictionaries. For example, even if only the separation rule at the beginning of the dictionary is applicable and the other rules have no applicability, all rules in the dictionary are checked for applicability.

For this reason, in the conventional word spelling-one-phonetic symbol conversion method,
There was a problem that there were a lot of useless searches (and the exception word processing was not performed efficiently).

Furthermore, if there is a word that has an exceptional pronunciation only when a certain affix is connected, if you search the exception word dictionary for the stem after affix separation processing, the stem from which the affix has been separated will no longer be an exception word. There was a problem in which dictionary searches failed.

Each time a suffix is separated from a spelling by the affix separation rule, the present invention searches for the existence of the separated word stem of that connection in the exception word processing unit, thereby speeding up exceptional word processing. The purpose of this invention is to provide an improved method for converting word spelling and phonetic symbol 2, which is improved so that the word spelling and phonetic symbol conversion can be performed without errors by preventing the failure of exceptional word dictionary searches due to the separation of extra affixes. .

[Means for solving problems]

The measures taken by the present invention will be explained with reference to FIG. FIG. 1 is a block diagram showing the basic configuration of the present invention.

In FIG. 1, reference numeral 10 denotes an exceptional word processing means for processing words with exceptional pronunciation, which includes an affix separation processing section 11 and an exceptional word processing section 12.

The affix separation processing unit 11 performs a process of individually separating affixes from words or stem spellings according to predetermined affix separation rules.

Every time an affix is separated from the spelling of a word or a word stem, the exception word processing unit 12 performs a process of searching whether or not the stem part from which the affix has been separated corresponds to an exception word.

[For production]

When the word spelling is input to the exception word processing means 10, the affix separation processing section 11 performs a process of separating one affix from the word spelling according to a predetermined affix separation rule.

The exception word processing unit 12 searches the stem part of a spelled word from which one affix has been separated to see if it corresponds to an exception word. When a search result corresponding to the exception word is obtained, the exception word processing is terminated. If the search fails, the affix separation processing unit 11 is notified and causes the next affix separation process to be performed.

In the same way, the exception word processing unit 12 performs processing to search whether the stem part from which the affix has been separated corresponds to an exception word each time one affix is separated from the word stem by the affix separation processing unit 11. When it is found that the word stem part corresponds to an exceptional word, the exceptional word is immediately terminated, and only when it does not correspond to an exceptional word, the affix separation processing unit 11 is made to perform the next affix separation process.

By doing the above, it is possible to eliminate the waste of checking the applicability of all affix separation rules as in the conventional method, and to speed up exceptional word processing. In addition, each time an affix is separated, a search process is performed to see if it corresponds to an exceptional word, so it is possible to prevent an extra affix from separating and failing to search for an exception word, and the word spelling and phonetic symbol conversion process is performed. can be performed without error.

〔Example〕

An embodiment of the present invention will be described with reference to FIGS. 2 and 3. FIG. 2 is a block diagram illustrating the configuration of an embodiment of the present invention, and FIG. 3 is a diagram illustrating a specific example of word spelling and phonetic symbol conversion processing according to the embodiment.

(A) Structure of Embodiment In FIG. 2, the exception word processing means 10 and the affix separation processing unit IL exception word processing section 12 are as described in FIG.

In the affix separation processing unit 11, an affix separation rule dictionary 111 stores affix separation rules for separating affixes from words or word stems.

Reference numeral 112 denotes an affix separation rule reading unit that performs a process of individually reading affix separation rules from the affix separation rule dictionary 111 in a predetermined order. Since the order of arrangement of various suffixes and prefixes is determined, the suffixes are read from the separation rule of the last suffix. If it is a prefix, it is read from the first prefix separation.

Reference numeral 113 denotes a rule end determination unit, which includes an affix separation rule dictionary 11
It is determined whether reading of all affix separation rules in 1 has been completed.

A rule application determination unit 114 performs processing to determine whether the read affix separation rule is applicable to the input word or word stem.

An affix separation processing unit 115 applies an affix separation rule to the input word or word stem when applicable, and performs affix separation processing.

In the exceptional word processing unit 12, 121 is an exceptional word dictionary, in which the spelling of a word with an exceptional pronunciation and the spelling of its phonetic symbol, or the spelling of a word to which an exceptional accent is added, and an accent added to the spelling. The spelling is stored in pairs.

Reference numeral 122 denotes a second exceptional word dictionary search unit that performs processing to search whether the spelling of the input word or word stem exists in the exceptional word dictionary 121.

Reference numeral 13 denotes a first exception word dictionary search unit, which performs processing to determine whether the spelling of the input word exists in the exception word dictionary 121.

Reference numeral 21 is an accent position estimation rule dictionary, which stores rules for adding an accent to the spelling of words.

Reference numeral 22 denotes an accent position estimating unit that refers to the accent position estimation rule dictionary 21 and performs a process of estimating an accent for an input word spelling or a spelling that is a combination of an unconverted affix spelling and a phonetic symbol.

A pronunciation rule dictionary 23 stores rules for converting accented spellings into phonetic symbols.

24 is a spelling and phonetic symbol converter, and a pronunciation rule dictionary 23
, and converts the input spelling with accent information into phonetic symbols.

(B) The spelling of the word read from the action sentence of the example is the exception word processing means 10
, the first exceptional word dictionary search unit 13 searches the exceptional word dictionary 121 to determine whether the entire spelling of the input word has an exceptional pronunciation or accent. The dictionary search is performed using, for example, a hash (I(ash)) function.

If the word spelling exists in the exception word dictionary 121, the character string of the pair is replaced with the input word spelling, and the exception word processing ends. In this case, the output character string is a phonetic symbol (hereinafter indicated by ■) or a spelling with accent information (hereinafter indicated by ■). In the case of the phonetic symbol ■, it is output as is, and in the case of the spelling ■ with accent information, it is input to the spelling-one-phonetic symbol converter 24.

If the spelling of the word does not exist in the exception word dictionary 121, the search fails, and the affix separation rule reading unit 112 is notified of this fact.

When the affix separation rule reading unit 112 receives a notification of search failure from the first exception word dictionary search unit 13, it reads the first affix separation rule from the affix separation rule dictionary 111 in a predetermined order, and ends the rule. Determination unit 113
send to

The rule end determination unit 113 determines whether all affix separation rules in the affix separation rule dictionary 111 have been read from the input affix separation rules. If all affix separation rules have been read, the input word is not an exception word, as will be clear from the explanation below, so the exception word processing is finished and the input word spelling (indicated by ◎ below) is determined. It is sent to the accent position estimating section 22.

If all the affix separation rules have not been read, the read affix separation rules are stored in the rule application determination unit 1.
14 to determine whether it is applicable to the input word spelling. If it is not possible to apply to the input word spelling,
The affix separation rule reading unit 112 is notified of this fact.

When the connection separation rule reading unit 112 receives this notification, it reads the next affix separation rule from the affix separation rule dictionary 111, and then the connection separation rule reading unit 112 reads out the next affix separation rule from the affix separation rule dictionary 111, and the rule end judgment unit 113 and the rule application judgment unit 1
14 and performs each of the above-mentioned processes.

When the rule application determining unit 114 determines that the affix separation rule is applicable to the spelling of the input word, the affix separation rule and the spelling of the input word are sent to the affix separation processing unit 115 to perform affix separation processing.

The affix separation processing section 115 performs affix separation processing on the word spelling according to the input affix separation rule, and sends it to the second exception word dictionary search section 122 .

The second exception word dictionary search unit 122 searches whether the generated word stem portion exists in the exception word dictionary 121. If the search fails, that is, the generated stem spelling is the exception word dictionary 121.
If the affix separation rule does not exist, the affix separation rule reading unit 112 is notified.

Upon receiving this notification, the affix separation rule reading unit 112 reads the next affix separation rule from the affix separation rule dictionary 111, and repeats each of the above-described processes.

If the search is successful, that is, if the generated stem spelling exists in the exceptional word dictionary 121, the spelling of the word stem is replaced with the character string of the pair, and the exceptional word processing is completed. In this case, the output character string contains unconverted affix spellings (hereinafter indicated by the ten phonetic symbol @) in the phonetic symbol of the root part, and it is necessary to determine the accent position of these. , is sent to the accent position estimation section 22.

Below, with reference to Figure 3, the word spelling r improνm
ents J with its phonetic symbol (impru niji me)
The case of converting to (nts) will be explained as an example.

In the word spelling r improvements J,
r imJ is a prefix, rmentJ and the ending r
sJ is a suffix, and r prove J is the original root. “prove J is an exceptional word in which the pronunciation of roJ is [U:], and only this r prove J is stored in the exceptional word dictionary 121.

Spelling of words read from sentences r improvme
ntS" is input to the exception word processing means 10, the first exception word dictionary search unit 13 searches the input word spelling r improv.
ments J as a whole is checked by searching the exceptional word dictionary 121 to see if it has an exceptional pronunciation or accent. Since rimprovments J does not exist in the exception word recess 121, the search fails, and the affix separation rule reading unit 112 is notified of this fact.

In the input word spelling rimprovments J, the affix separation rules for the suffixes rsJ, rmentJ and the prefix rimJ are applicable.

First, the affix separation rule for the final suffix rsJ,
For example, “φ instant * / C (CC) 3 shown in Figure 3 (bl)
*" is applied. In each affix separation rule in Figure 3, ■ is a vowel, C is a consonant, and * at both ends indicates the ending or beginning of a word. The affix separation rule ■ shown in Figure 3 (bl) is 1-3
If there are four consonants followed by rsJ, then rs
The rule is to insert an affix separator * before JO.

By applying this affix separation rule, the final suffix rsJ is separated, as shown in Figure 3 (C1).
improvement*, s*'' are generated.

The stem part surrounded by * in the generated character string [impr
Since ovments J does not exist in the exception word dictionary 121, the second exception word dictionary search unit 122 fails in the search and notifies the affix separation rule reading unit 112 to read the next affix separation rule.

The next applicable rule is the affix separation rule ■ that separates the suffix r ment J. For example, as shown in Figure 3 (dl), “φ→e *, /VC(CC) ment
*”.This affix separation rule ■ is “vowel + 1 to 2 consonants + ment*”, then “men
The rule is to insert e and * before t''.

By applying this affix separation rule ■, the suffix r
Men t J was separated, and as shown in Figure 3 tel “
*improve*, ment*, s*” are generated.

Since the word stem molecule improve J in the generated character string does not exist in the exception word dictionary 121, the second exception word punctuation search unit 122 fails in the search, and the affix separation rule reading unit 1
12 to read the next affix separation rule.

The next applicable rule is the affix separation rule ■ which separates the prefix r imJ. For example, as shown in Figure 3(f), “
φ→, */* im C(CC) V (V)C
(CC)V”. This affix separation rule ■ is given by
“*im + 1 to 3 consonants + 1 to 2 vowels + 1
The rule is that if "*im" is in front of a string of two or more syllables, an affix separator * is inserted after "im". .

By applying this affix separation rule■, the prefix ``i''
mj is separated, and as shown in Figure 3(g), “*im,
*prOVe*, ment*, s*'' are generated. The above-mentioned affix separation rules are disclosed in detail in, for example, Japanese Patent Application No. 60-271261 and Japanese Patent Application No. 61-122009.

The stem part surrounded by * in the generated character string "prov
e J exists in the exceptional word dictionary 121 because the vowel "0" is pronounced as [U;] in an exceptional manner.

When the second exceptional word dictionary search unit 122 succeeds in searching for r prove J, the second exceptional word dictionary search unit 122 converts the word stem molecule prove J into the phonetic symbol (pru:v) read from the exceptional word dictionary 121, as shown in FIG. 3(i). Replace.

“*im, *(p ru
:V) *, ment*, s*” are combinations of unconverted affix spellings and phonetic symbols, and the accents are undetermined, so they are sent to the accent position estimating unit 22 as spelling ten phonetic symbols ■. .

The accent position estimating unit 22 is connected to the rule end determining unit 113.
Word spelling input from ◎ or second exception dictionary search section 1
For the spelling ten phonetic symbols ■ input from 22, accent information is added according to the rules read from the accent position estimation rule dictionary 21, and the result is sent to the spelling one phonetic symbol conversion section 24.

The spelling 1 phonetic symbol conversion unit 24 refers to the pronunciation rule dictionary 23 and performs a process of converting the spelling with accent information inputted from the accent position estimation unit 22 or the first exception word dictionary search unit 13 into a phonetic symbol. .

In the manner described above, the spelling of an input word can be converted into the corresponding phonetic symbol.

Next, the word spelling-phonetic symbol conversion method of the present invention and the fourth
The exception word processing efficiency compared to the conventional method shown in the figure will be explained.

The types of input words are (exceptional words in which the al affix is not connected, (
There are four possible types of words: exceptional words connected with the bl affix, (C) regular words not connected with the affix, and regular words connected with the dl affix.For each word, the time required for exceptional word processing is In the case of the conventional method, Ta1, Tbl
, Tcl, Tdl, the case of the method of the present invention is Ta2.
Tb2. Tc2. Let it be Ta2.

(i) In the case of an exception word with no connected affixes, both methods have an exception word processing section first, so the processing time is equal to the search time of the exception word dictionary, and the same dictionary and the same search method are used. As long as Ta1=1Ta2.

i) For regular words with no connected affixes, all affix separation rules and spellings will be matched, and as long as the same affix separation rule dictionary and the same matching algorithm are used, the processing time will be Both equations are equal, resulting in Tc1-Tc2.

iii) For regular words connected by affixes, Tdl
= Time required to match all rules and spellings +
Extraneous word dictionary search time Td2 = time required for matching all rules and spellings x extraneous word dictionary search time x number of affixes. However, if a high-speed algorithm is used for exceptional word dictionary searches, as mentioned earlier, the number of affixes connected to one word is at most seven, and the number of exceptional word dictionary searches is about seven times. Therefore, Tdl and Ta2 are considered to be approximately equal.

iv) In the case of an exceptional word with connected affixes, Tb1 = Time required to match all rules and spellings + Extra-lateral language dictionary search time Tb2 = Time required to match spellings with rules up to the applicable rule This is the time required to search the dictionary for ten exceptions. If the rule applicable to the input spelling is at the end of the rule dictionary, Tb1=Tb2, but otherwise Tbl>Tb2.

Therefore, Ta 1 +Tb 1 +Tc 1 +Td
1≧Ta 2 + Tb 2 + TC2 + Td 2,
The method of the present invention can perform exceptional word processing faster than the conventional method.

Although the embodiment of converting the spelling of an English word into a phonetic symbol has been described above, the present invention is also applicable to a word spelling-one-phonetic symbol conversion method for languages other than English.

〔Effect of the invention〕

As explained above, according to the present invention, the following effects can be obtained.

(b) Every time an affix is separated from a word or the spelling of a word stem, a search is made to see if the stem part has an exceptional pronunciation or accent, and if the search is successful, the separate external processing ends immediately, which is better than the conventional method. can also speed up exceptional word processing.

(b) Each time an affix is separated, a search process is performed to see if it corresponds to an exceptional word, so it is possible to prevent an extra affix from separating and failing to search for an exception word, and to convert the spelling of a word into a phonetic symbol. Processing can be performed without error.

[Brief explanation of the drawing]

FIG. 1 is an explanatory diagram of the basic configuration of the present invention, FIG. 2 is an explanatory diagram of the configuration of an embodiment of the present invention, and FIG. 3 is an explanatory diagram of a specific example of word spelling and phonetic symbol conversion processing according to the same embodiment. , FIG. 4 is an explanatory diagram of the conventional word spelling-one-phonetic symbol conversion method. 1 and 2, 10... exceptional word processing means, 11... affix separation processing section, 12... exceptional word processing section, 13... first exceptional word dictionary search section, 21.・Accent position estimation rule dictionary, 2
2...Accent position estimation unit, 23...Pronunciation rule dictionary, 25...Spelling and phonetic symbol conversion unit.

Claims (1)

[Scope of Claims] A word spelling-phonetic symbol conversion method for converting an arbitrary sentence into speech, comprising an exceptional word processing means (10) for processing the word spelling of an exceptional word with an exceptional pronunciation, The exceptional word processing means (10) includes: (a) an affix separation processing unit (11) that individually separates affixes from the spelling of a word or word stem according to a predetermined affix separation rule;
(b) Each time an affix is separated from the spelling of a word or a word stem, an exception word processing unit (12) is provided which searches whether the stem part from which the affix has been separated corresponds to an exception word. Features a word spelling-phonetic symbol conversion method.