JPS63155229A - Conversion system for word processor - Google Patents

Abstract

PURPOSE:To eliminate the manual operation by performing character conversion or line conversion by the time length of a voice. CONSTITUTION:A voice signal from a microphone 1 is inputted to a voice recognition part 3. A time length signal S of the voice is sent out of the voice recognition part 3 to a template switching part 4. The template switching part 4 judges whether or not the voice data is a conversion command or not from the difference between the time length of the voicing time a syllable and the time length of the voicing time of a word and switches templates. A selected single-syllabry 5 or word templete 6 is supplied to the voice recognition part 3 through the template switching part 4 to identify the word or generate a conversion command.

Description

[Detailed Description of the Invention] [Summary] The present invention provides a conversion method in a voice input word processor that enables character conversion, line conversion, etc. to be performed by voice instead of using a keyboard by identifying the time length of human voice. be.

[Industrial application field]

The present invention relates to a word processor, and more particularly to a conversion method for a word processor that allows voice input.

[Prior Art and Problems to be Solved by the Invention] In recent years, with the development of speech recognition technology, voice input has become commonplace in place of human input using the keys of a word processor.

In this case, the speech recognition device that is normally used uses a device that stores linguistic information of various dimensions such as phonemes, words, and syntax to recognize the phonemes by checking the degree of similarity with the input speech based on a standard phoneme pattern. . The device is equipped with a so-called word dictionary (template), which includes katakana, hiragana,
Character strings such as kanji are stored and referenced for word recognition.

FIG. 4 shows an example of the main part of the conventional device described above.

In FIG. 4, 1 is a microphone for voice input, 2 is a keyboard for inputting data, commands, etc., 3 is a voice recognition section, 4 is a template switching section, 5 is a monosyllabic template for text data, 6 is a word template for commands, 7 is a text processing section, and 8 is a display device (CRT).

In such a configuration, if you want to input "speech", utter the monosyllables "o", "n", "se", and "i" as shown in Figure 5, and the voice will be recognized through microphone 1. The information is input to section 3. The speech recognition unit 3 recognizes “o”, “n”, “se”, and “i” while checking against the monosyllable template 5, and inputs the result to the sentence processing unit 7. after that,
As shown in FIG. 5, when the command key on the keyboard 2 is pressed (turned ON), the template switching unit 4 switches to the command word template 6, matches the input voice of "Henkan", and instructs the character conversion process. After recognizing this, an instruction for conversion processing is sent to the text processing section 7. Thereby, the "audio" input data can be displayed on the display device 8.

In this way, conventionally, when converting voice input, it was done by pressing the command key, and even though voice input is possible for the data part, the command key is still used for control parts such as character conversion instructions. This made manual operations complicated.

[Means and effects for solving the problems] The present invention provides a conversion method for a word processor that solves the above-mentioned problems. Focusing on the difference, the system allows switching between audio data and conversion commands using only voice without keyboard operations, and the method is as follows:
Equipped with a voice recognition unit that detects the time length of input voice and compares it with a reference value, and when the time length of the input voice exceeds the reference value, it recognizes the input voice as a conversion command and converts characters or lines. etc., and when the time length does not exceed the reference value, it is recognized as text data and displayed.

〔Example〕

FIG. 1 is a basic configuration diagram of a word processor conversion method according to the present invention. As is clear from the figure, the basic device configuration is the same as that of the conventional one, but in the present invention, the speech time length signal S is sent from the speech recognition section 3 to the template switching section 4, and the utterance time of a single syllable is The template is switched by detecting the difference between the time length and the time length of the utterance of the word, determining whether it is voice data or a conversion command. Furthermore, the fourth
Although the illustrated keyboard 2 is necessary for other functions, illustration thereof is omitted.

A more detailed explanation will be given with reference to FIG. In a word processor using voice input, for example, "voice"
When entering the word ``o'', ``n'', ``se'',
When you utter "i" in single syllables, the speech recognition unit 3 compares it with the monosyllable template 5 and utters ``o'', ``n'', ``se'', etc.
I recognize that. In this case, the speech time can be seen to be almost the same for each monosyllable, as shown by t to t4. Note that the vertical axis in FIG. 2 represents the loudness of the sound, and the horizontal axis represents the time.

Next, when "Henkan" is uttered without separation, the utterance time t is longer than when it is a single syllable, so the speech recognition unit 3 detects this difference in time and converts the speech duration signal S into a template. The switching unit 4 switches to the word template 6 for the sending command, performs a check, and after recognizing that it is a "conversion process", sends a conversion command C and notifies the text processing unit 7.

Generally, the time length of monosyllable utterance is about 230 m5ec, and the length of time uttered of a word is about 1 sec. However, there are individual differences between speakers, and if it cannot be determined generally, it is possible to use a temporarily long voice as a command based on the distribution of the time length of the input voice. It is also possible.

FIG. 3 is a control 'flI flowchart of the conversion method according to the present invention. In Figure 3, the sound is transmitted to microphone 1.
is input to the speech recognition unit 3 through the voice recognition unit 3 (step 1). In the speech recognition unit 3, a comparison means (not shown) for comparing the time length of the input speech determines whether the input speech is a monosyllable or a word. This judgment criterion is, for example, compared with a standard time length, and when the human voice is shorter than the standard time length, it is monosyllable,
If it is long, it is determined that it is a word, and a voice duration signal S is output (step 2). Then, in the template switching unit 4, in the case of a monosyllable, a monosyllabic template 5 is selected (
Step 3), the word is identified in the speech recognition unit 3 with reference to the word dictionary (step 4), and then passed through the sentence processing unit 7 (
Step 5) Displayed (Step 6).

On the other hand, in the case of a word, the template switching unit 4 selects the word template 6 (step 7). It is recognized as a control command in the speech recognition section 3 with reference to the control command word dictionary (step 8), and a conversion command C is created in the control command generation section (not shown) (
Step 9), character conversion is performed in the text processing unit 7 (
Step 5) is then displayed (Step 6). Note that not only character conversion but also line conversion can be similarly performed by determining the time length.

〔Effect of the invention〕

As explained above, according to the present invention, when converting characters or lines in a voice input word processor, it is possible to perform character conversion or line conversion based on the duration of the voice rather than using command keys, thereby eliminating the complexity of manual operations. .

[Brief explanation of the drawing]

Fig. 1 is a block diagram of a word processor conversion method according to the present invention, Fig. 2 is a diagram explaining character conversion, Fig. 3 is a flowchart showing a character conversion method according to the present invention, and Fig. 4 is a configuration of a conventional method. 1 and 5 are diagrams illustrating conventional character conversion. (Explanation of symbols) l...Microphone 2...Keyboard 3...Speech recognition section 4...Template switching section 5...Monosyllabic template 6...Word template 7...Sentence processing section 8 ... Display device Configuration diagram of the character conversion method according to the present invention $1 Figure Detailed explanation of the non-invention Figure 2 Flowchart of the method according to the present invention Figure 3

Claims (1)

[Claims] 1. A conversion method for a voice input type word processor in which text data is input using the input voice when the text data is read aloud. When the duration of the audio exceeds the standard value, the input audio is recognized as a conversion command and converted into text or lines, and when the duration does not exceed the standard, it is converted into text data. A conversion method for a voice input word processor characterized by recognition and display.