JPH01154100A - Voice recognition result confirmation method - 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] Gaihobu! The present invention relates to a method for checking speech recognition results, and more particularly to a method for displaying and selecting recognition results of a speech recognition device or a word processor using speech input.

Prior Art In the conventional method for displaying and selecting recognition results in a word processor using voice input, if the correct answer (desired recognition result) is not obtained as the first candidate, the user presses a predetermined "next candidate key". The user must select the second and lower candidates sequentially or enter the information again.

In particular, even if the desired result is displayed each time the "next candidate key" is pressed and the user repeatedly checks, if the correct answer is not included in the candidates, the user must re-enter, which places a burden on the user. increase

Purpose The present invention was made in view of the above-mentioned circumstances.
In particular, in a voice input document creation device such as a word processor that uses voice input, selection and confirmation operations of recognition results can be carried out without burdening the user. This was done with the purpose of providing a voice recognition result confirmation method that has easier-to-use functions.

Configuration In order to achieve the above object, the present invention includes a microphone for inputting voice, a voice recognition section for recognizing the input voice, a recognition result confirmation section for holding and confirming the recognition result, and a recognition result confirmation section for retaining and confirming the recognition result. It consists of a display that displays the results, a keyboard that corrects the recognition results, and a data processing unit that performs language processing such as kana-kanji conversion on the corrected results, and simultaneously processes up to the nth (n is an integer) candidate of the recognition results. The present invention is characterized in that it has a display means, and selects the 0th candidate from the displayed candidates by using keys such as a numeric keypad. Hereinafter, the present invention will be explained based on examples.

FIG. 1 is a block diagram of a system configuration for explaining an embodiment of the present invention. In FIG. 1, 1 is a microphone;
2 is a voice recognition unit, 3 is a recognition result confirmation unit, 4 is a data processing unit, 5 is a display, and 6 is a keyboard.Speech input from microphone 1 is analyzed and collated by voice recognition unit 2, and then recognized. The results are held in the result confirmation section 3 and displayed on the display 5 at the same time. If the recognition result is correct, the data processing unit 4 immediately performs language processing such as kana-kanji conversion. However, if the recognition result contains an error, it is necessary to perform appropriate correction processing using the keyboard 6 and then send the correction result to the data processing section 4.

In addition to the correction process using the keyboard 6 described above, it is also possible to correct the correction part by performing voice input again. Similarly, it is also possible to modify the results of data processing in the data processing section 4 again using the keyboard 6.

FIG. 2 is a diagram for explaining the display form of the recognition results and the correction method, and 51 in FIG. 2(a) indicates an example of the recognition results displayed on the display 5 of FIG.

Now, when the input voice is "Onsei".

The recognition results are displayed on the display 5 as syllable candidates. That is, for the first syllable "o", it is "ha", "ko", "oJ", and for the second syllable "n", it is "u", "n", etc. Note that the number of candidates to be displayed is appropriately selected based on the degree of similarity during pattern matching in the recognition unit 2.

61 in FIG. 2(d) is the numeric keypad in the keyboard 6;
Reference numeral 62 in Figure (e) represents a cursor movement key, and upon viewing the result display 51, the user selects the recognition order of each syllable in order using the numeric keypad. In this example, the recognition ranking of each syllable is 3rd place, 2nd place
1st place, 2nd place, so the numeric keypad is 3 → 2 → 1 → 2
and press.

To do this, the cursor position is moved to the first syllable using the cursor movement key 62 before the correction operation, and then the correction is performed.

However, when multiple candidates are displayed as shown in 51, it is also possible to automatically move the cursor position to the first syllable (uncorrected syllable) position (A) before performing the correction operation (i.e., multiple candidates). It is only necessary to detect that a candidate is displayed).

By successively pressing the numeric keypad in this way, the display 51 changes to the display 52 in FIG. 2(b) (syllables other than the correct answer are erased). The correction result 52 is converted into kanji by the data processing unit 4, and the conversion result 53 (Fig. 2 (C
)) is obtained.

If the first recognition result is correct, such as the third syllable "se" at 51 in FIG. 2(a), it is possible to add a function that does not require pressing the numeric keypad "1". That is, the third
If the fourth syllable is input after the syllable is displayed, the first recognition result for the third syllable may be automatically selected.

However, when this function is selected, each syllable must be corrected one syllable at a time, and as mentioned above, the correction operation that involves moving the cursor after entering a single phrase (or ' is a word) is a different mode. It is necessary to do it.

Furthermore, it goes without saying that the results of kana-kanji conversion in the data processing unit 4 shown in FIG. 1 can be corrected using the keyboard 6 (this is one of the basic functions of a word processor). If the data processing unit 4 inserts a word delimiter code or phrase delimiter code into the syllable recognition result by inputting it from the keyboard 6, it becomes possible to improve the conversion accuracy of kana-kanji conversion.

Although the method for correcting and confirming recognition results based on monosyllable input has been described above as an embodiment of the present invention, the present invention is not limited to this.

Fixes related to word-by-word and even continuous speech input.

It is also possible to apply the same method to the confirmation method.

However, in this case, word candidates etc. are substituted for the syllable candidates shown in FIG.

Effects As is clear from the above explanation, according to the present invention, by displaying a plurality of speech recognition results at the same time and selecting the correct candidate using the numeric keypad, confirmation and correction operations for the results are simplified. The burden on the user is reduced.

[Brief explanation of the drawing]

FIG. 1 is a system configuration block diagram for explaining an embodiment of the present invention, and FIG. 2 is a diagram for explaining a display form of recognition results and a correction method. 1...Microphone, 2...Speech recognition unit, 3...
- Recognition result confirmation unit, 4... data processing unit, 5... display, 6... keyboard.

Claims (3)

[Claims] (1) A microphone that inputs voice, a voice recognition unit that recognizes the input voice, a recognition result confirmation unit that holds and confirms the recognized results, a display that displays the recognition results, and a display that corrects the recognition results. It consists of a keyboard and a data processing unit that performs language processing such as kana-kanji conversion on the corrected results, and has means for simultaneously displaying up to the nth (n is an integer) candidate of the recognition results. A voice recognition result confirmation method characterized by selecting the nth candidate from the candidates using keys such as a numeric keypad. (2) When a plurality of voice recognition results are displayed as candidates, the first candidate is automatically selected by continuously inputting the next voice. The voice recognition result confirmation method described. (3) When multiple voice recognition results are displayed as candidates,
The voice recognition result confirmation method according to claim 1, wherein the cursor is automatically moved to the first plural candidate display position and waits for a correction input.