JPH0766276B2 - Speech input word processor for monosyllabic input - Google Patents

Description

TECHNICAL FIELD The present invention relates to a word storage device used in a voice input word processor for single syllable input.

(Technical background) In a speech input word processor or the like using a single syllable input, normally, a uttered syllable cannot be uniquely recognized, and therefore a word composed of a collection of syllables cannot be uniquely recognized. Therefore, when one word is uttered in a single syllable, a plurality of words are recognized as candidates.

In addition, when 2 to 5 syllables are set as the candidate syllables for one syllable, there is a drawback that the uttered syllables are not in the candidate syllables, and thus the words that collect them may not be correctly recognized. .

Furthermore, even if all the syllables of a word composed of multiple syllables include the uttered syllable as a candidate syllable, the word stored in the word storage device is a Japanese syllabary like a general Japanese dictionary. If the words are arranged in order, for example, if the number of characters that make up a certain word is M and the fixed number of candidate syllables for each syllable is N, it is necessary to extract the word N ^M times, which increases processing speed. It also had the drawback of falling significantly.

(Object of the Invention) An object of the present invention is to solve the above two drawbacks, and it is possible to extract a word including a syllable that is not in a candidate syllable as a candidate word, and further to determine a candidate word in a shorter processing time. A storage device is provided.

(Structure of the Invention) The present invention includes (a) a syllable length table that stores the address of a vowel sequence code table that stores a combination of vowels that can be configured by the syllable length for each recognized syllable length, and (b) preset. A vowel sequence code table that stores a combination of vowels that can be configured with a specified syllable length, the number of words that have the combination of vowels, and the address of a word table that stores that word; and (c) the heading of the word. ,
It has a word table that stores the control information of that word,
The table of (b) is accessed with the address of the table of (a), and the table of (c) is accessed with an address corresponding to the vowel code of the table of (b) that is called as a result of the access. Is a voice input word processor for monosyllabic input, which is characterized in that the control information is read and the read word is output, which will be described in detail below with reference to embodiments.

(Embodiment) FIG. 1 shows the inside of a word storage device according to an embodiment of the present invention. In FIG. 1, 1 is a syllable length table, 2 is a vowel sequence code table, and 3 is a word table.

Vowel sequence code table 2 that can store 20 16-bit long addresses in the syllable length table 1 and stores combinations of vowels that can be configured by each syllable length, with the minimum syllable length being 1 syllable and the maximum 20 syllables. Addresses 4 of are stored respectively.

The vowel sequence code table 2 includes a vowel code 5 for each syllable length, the number 6 of words having the same vowel code as the vowel code, and an address 7 to the word table 3 in which the word is stored. It is stored. As the vowel chord, as shown in Fig. 2, 1.
Correspond to "a" stage 2. "c" stage 3. "e" stage 4. "o" stage 5. "n" 6 They are arranged in an ascending order for easy search.

The word table 3 stores a word heading 8 in Kana characters coded in 8 bits and control information 9 about the word, which are arranged in alphabetical order.

Next, a case where the syllable length is 3 will be described as an example with reference to FIGS. 3 and 4. In the syllable length table, the syllable length is 3
The address corresponding to is stored in the address of the vowel sequence code table that stores the combination of vowels formed by the syllable, and is accessed as indicated by the arrow in FIG.

The vowel sequence code table has 5 × 6 ² entries, the vowel code at the beginning is “Ah”, and the final vowel code is “On”. When expressed by the numbers in FIG. 2 described above, they are “111” and “565”, respectively. Behind the vowel code, the number of words each having the vowel code is stored. For example, in FIG.
It shows that it is an individual. Further, following the number of words, the address of the word table storing the word having the vowel code is stored, and the address is accessed as indicated by the arrow.

Then, the word table is accessed, and the words in the word table are output to a display unit (not shown) to identify the input word, and one of the output words is selected.

FIG. 3 shows a word table in which words having the vowel code "aahh" are stored, and "acasa", "acama", ... "Warawa" and the like are stored. In addition, part-of-speech information is encoded and stored as control information for each word. Part of the part-of-speech code is shown in FIG. From Fig. 3 and Fig. 3, "Akasa" is a noun, "Akama" is a person's name,
It can be seen that "Warawa" is a noun. This control information is used for checking the connectivity of the voice-input character string in the voice input word processor or the like using the present invention.

As described above, since words having the same vowel code are stored in one place for each syllable length unit and accessed, there is no need to access the word storage area intermittently as in the conventional Iueo order. The number of accesses is reduced and high-speed processing becomes possible.

(Effects of the Invention) As described above with reference to the embodiments, according to the present invention, words having the same vowel code for each syllable length are collectively stored in one place on the word storage device. Since it is not necessary to access the word storage area in an irregular manner, such as in the case of AIUEO, the number of word accesses can be reduced and high-speed processing can be performed. Further, even in the conventional voice recognition technology, there is almost no error in recognizing a vowel code, so even if there is an erroneously recognized syllable, the correct word is included in the word group having the same vowel code. Words can also be included in the candidate words, and recognition errors can be automatically corrected.

In addition, by outputting a word group having the same vowel code to a display device or the like, it becomes possible for the operator to select a desired word, which increases convenience in selecting a correct input word as a word processor.

[Brief description of drawings]

FIG. 1 is a diagram showing an embodiment of the present invention, an explanatory diagram for explaining a table relation of a word storage device, FIG. 2 is an explanatory diagram for explaining correspondence between a vocal sound and a vowel code, and FIG. 3 is a syllable length. 4 is an explanatory view for explaining an embodiment in the case of 3 and FIG. 4 is an explanatory view showing a part thereof for explaining the part-of-speech code. 1 ... syllable length table, 2 ... vowel sequence code table, 3 ... word table, 4 ... vowel sequence code table address, 5 ... vowel code, 6 ... number of words,
7 ... Address of word table, 8 ... Heading, 9 ...
Control information.

Claims (1)

[Claims] 1. A syllable length table storing the address of a vowel sequence code table storing a combination of vowels that can be formed by the syllable length for each recognized syllable length, and (b) a preset syllable length. A vowel sequence code table storing the combinations of vowels, the number of words having the combination of vowels, and the address of the word table storing the words; and (c) the heading of the word and the word A vocabulary table storing control information, and accessing the vowel sequence code table at an address of the syllable length table, and a word table at an address corresponding to a vowel code of the vowel sequence code table called as a result of the access. To read a word and control information, and output the read word. Audio input word processor.