JPS59186041A - Voice document preparing device - Google Patents

Abstract

PURPOSE:To shorten time for inputting, facilitate ''kana-kanji'' (Japanese syllabary-Chinese character) conversion and improve processing speed by inputting part corresponding to kanji in inputted sentence continuously in two syllables and inputting parts corresponding to kana in monosyllable. CONSTITUTION:A voice recognizing section 3 converts inputted voice to corresponding kana letter information and stores in a memory 30. A ''kana-kanji'' converting section 4 converts stored kana letter information to writings mixed with ''kanji'' basing on ''kanji-kanji'' word dictionary 41. In this case, monosyllabic parts are written in kana, and parts of two continuous syllables are written in ''kanji''. Thus, as the parts corresponding to ''kanji'' are inputted in two continuous syllables, number of punctuation is reduced compared with the case where inputting is made by one syllable and time for inputting can be reduced. Further, as ''kanji'' parts and kana parts are inputted separately, ''kana-kanji'' conversion becomes easy, time for processing is shortened and processing speed can be increased.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an audio document creation device that can be applied to audio typewriters and the like that create documents through audio input.

Support 3 Conventionally, an audio document creation device that inputs monosyllable units has been announced. However, in such an audio document creation device, since the whole sentence must be pronounced one by one by cutting the IK, it takes a long input time. Furthermore, since information regarding the character type is not included even though speech is used, the amount of processing in the kana-kanji conversion unit is almost the same as that for inputting kana characters, so there is little merit in using speech input.

Therefore, a voice word processor using input in units of single j/i may be considered, but the number of blanks in the dictionary that describes the voice features for matching or searching is equal to the number of single, ;t1, which is extremely large, making it difficult to put into practical use. Have difficulty.

Rule-1 In view of these circumstances, the present invention reduces the input time by inputting, in principle, the part corresponding to kanji in the input sentence as two consecutive syllables, and the part corresponding to kana as a single syllable. It is an object of the present invention to provide an audio document creation device which can shorten the number of words and at the same time easily perform kana-bakuji conversion processing, thereby improving the overall processing speed.

Mustache-1 The configuration of the present invention will be described below based on one embodiment.

FIG. 1 is a block diagram of an embodiment of the present invention, in which 1 is a voice input section, 2 is a key input section, 3 is a voice recognition section capable of recognizing two syllables in a row, and 3o is a storage unit for storing voice recognition results. 4 is a kana-kanji conversion unit, 40 is a storage unit for the kana-kanji conversion results, 41 is a kanji/kanji knee 1 dictionary, 5 is an output editing unit, 5o is a display unit, 6 is an external output unit, 7 is a control unit Department.

FIG. 2 shows a part of the contents of the kanji/kanji word dictionary 41.

Figure 3 shows an example of a keyboard used in the keypad section 2, where keys 2a are punctuation mark keys, 2b are kanji section designation keys, and 2c are
is the other key.

Voice input using such a configuration will be explained using the following example sentence.

<Example sentence> ``There are many recent reasons.'' When inputting this example sentence by voice, the operator manually puts a break in the utterance at the / position as shown below.

When there is a voice input such as say/kin/ame/ga/o-o/here, the sound j6 recognition unit 3 converts it into kana character information corresponding to the input voice and stores it in the storage device 30. to be memorized. The kana-kanji conversion unit 4 has means for determining whether or not there is a connection between parts of speech or between words.

A kanji character string is searched for to convert the stored kana character information into kanji kana characters based on the kanji/ambient language dictionary 41. In this case, there is information that the part where two syllables occur consecutively will be written in kanji! j-, this search process is basically limited to searching for the corresponding kanji part 1. In this example, we know that the first three characters are kanji, so we use the kana-kanji converter 4 to
The dictionary is searched using the longest match principle. Dictionary 4 is the second
As shown in the figure, it has readings, corresponding notations, and part-of-speech information. In this case, there is no kanji string corresponding to "Saisaname" in the dictionary, so "recently" and "recent" and ""Bacteria" remains as a search result candidate,
After being processed by the output editing section 5, it is displayed on the display section 50.

The operator selects an appropriate kanji string from among the displayed homophones. The same goes for "ame" and "oh".

Of course, it is possible to reduce the burden on the operator by incorporating a homophone selection function into the kana-kanji converter in advance. By manually inputting two consecutive syllables like this, the number of utterance breaks is reduced compared to the case where each syllable is input separately. ), this reduces the input time, and since the kanji part and the kana part are input separately, the algorithm for processing kana-kanji and The size of the dictionary used can be reduced, and processing time can be shortened.

Next, we will explain the processing of kanji with a reading of 1 character or 3 characters.

<Example sentence> ``I will appear.'' Input in this case is performed as follows using the voice and the kanji section designation key 2b (Fig. 3).

Wata key 2b comb/ga/key 2b ma/su If the voice input immediately before the kanji section designation key is two consecutive syllables, the voice immediately after the key means that it represents the reading of the rotating kanji. Therefore 27′? The ``kushij'' in the clause represents the reading of the same kanji that follows ``wata'', and the kanji that is read as ``watakushi'' is searched for, and ``watashi'' is displayed. Furthermore, if the voice input immediately before the Faculty of Commerce designation key is a monosyllable, it will represent a single kanji character such as J7t. In this way, it is possible to handle kanji of all reading lengths with just a single kanji part specification key. In this embodiment, the kanji part specification key 2b is used, but of course, by preparing a voice code, 77)! It is easy to configure i to play the same role as the Faculty of Commerce designation key 2b.

By the way, some kanji have a special reading as X/A-kyo. For example, the phrase ``明１'' can be read separately as ``miyo'' and ``nichi,'' but it can also be read as ``ashita'' as a whole. Therefore, the dictionary also includes the reading ``Ashita'' (one-letter notation) ``Tomorrow'', so that when inputting, it is possible to input the foot key 2b.

The processing for the kanji part is as follows. Note that the particles ``wa'', ``he'', and ``wo'' do not require any special processing as they can be read literally as I7.

“wo” is pronounced rWOJ.

Regarding the input of punctuation marks, in this embodiment, the punctuation marks are input using the dot keys 2a, but it is obvious that they can be input manually by providing a two-syllable phonetic code.

Note that the present invention can also be extended to continuous speech of three or more syllables. However, in many cases, the reading of kanji is 2.
In the case of two syllables, it is possible to eliminate the influence of the subsequent consonant on the consonant "n", the consonant "tsu", etc., and the problem of recognition processing caused by so-called articulatory combination can be simplified.

Enemy--1 As described below, in the audio document creation device according to the present invention.

(a) The input time can be shortened compared to inputting each single syllable.

(b) Since the Kanji part and the Kana part of the input sentence are clearly distinguished and input, the algorithm for kana-kanji processing and the capacity of the dictionary used therein can be reduced, and the processing time can be shortened.

(C) The burden on the operator can be reduced.

There are advantages such as

[Brief explanation of drawings]

FIG. 1 is a block diagram of an embodiment of the present invention, FIG. 2 is a diagram showing part of the contents of a Kanji/Kanji language dictionary, and FIG. 3 is a diagram showing a keyboard used for manual keying. 1...Voice input section, 2...Key human power section, 3...Voice recognition section, 4...Kana-kanji variant! ! :! Section, 5... Output editing section, 6... External output section, 7... Control section, 30
．． 40...Storage device, 41...Kanji/kanji language dictionary,
50...Display device. 1. ,′・,2 (J・,3i′;J

Claims (1)

[Claims] a voice input means, a key human power means, a voice recognition unit that recognizes continuous human voice of up to two syllables and converts it into kana character information corresponding to the human voice, a means for accumulating and storing the kana character information;
It has a kanji/kanji language dictionary that has notation and part-of-speech information corresponding to the reading, a means for determining whether or not connections can be made between parts of speech, and 374 words, and converts kana character information into kanji-kana mixed sentences based on the dictionary. The Kana-Kanji conversion unit that converts the Kanji/Kana-Mixed sentences, and the output editing unit that edits the Kanji/Kana-mixed sentences, as a general rule, input the part of the input sentence that corresponds to the Kanji in the 2-sound ffj tsuzuki "Kr voice. An audio document creation device characterized by inputting the pronunciation of a part corresponding to a kana in monosyllables.