JPH06130989A - Voice recognizing device - Google Patents

Abstract

PURPOSE:To provide the voice recognizing device for selecting a correct candidate out of plural candidates with simple operations. CONSTITUTION:The plural recognized results are outputted concerning a single syllable. Concerning the recognized result of a voice inputted from a microphone 1, a single syllable recognizing device 2 performs language processing by using a language processor 3, outputs the result while adding a score expressing certanity for the unit of a clause and displays the clause candidates on a display part 5 together with symbols for distinguishing the candidates in the score order of certanity. When the symbol displayed parallelly to the candidate for specifying the displayed candidate is voiced, this is recognized by the voice recognizing device 2 controlled by a controller 4.

Description

Detailed Description of the Invention

[0001]

TECHNICAL FIELD The present invention relates to a speech recognition device, and more particularly,
Japanese input device for monosyllabic recognition.

[0002]

2. Description of the Related Art Japanese input by voice is extremely natural as a means for inputting documents, and in general, monosyllabic recognition has attracted attention because it requires a smaller voice dictionary to be stored than word recognition. However, as shown in JP-A-3-221999, single syllable recognition generally has a low recognition rate, and it is necessary to output a plurality of candidates as a recognition result. Therefore, if the bunsetsu length becomes long, the number of bunsetsu candidates becomes extremely large even if language processing is performed, and there is a problem in that a correct candidate is selected from many candidates without burdening the user.

[0003]

[Purpose] As described above, inputting Japanese by voice is extremely natural as a means for inputting documents, and in particular, monosyllabic recognition has attracted attention because it requires a smaller voice dictionary to be stored than word recognition. However, the recognition rate of monosyllabic recognition is low and it is necessary to output multiple candidates as the recognition result. Therefore, if the bunsetsu length becomes long, the number of bunsetsu candidates becomes extremely large even if language processing is performed, and there is a problem in that a correct candidate is selected from many candidates without burdening the user. Therefore, an object of the present invention is to provide a voice recognition device that selects a correct candidate from a plurality of candidates by a simple operation.

[0004]

In order to achieve the above object, the present invention provides (1)
A single syllable recognition device that outputs a plurality of recognition results for one single syllable, a language processing device that performs a language process on the recognition result and outputs a score indicating the certainty of each syllable, and the order of certainty scores. A display device that displays phrase candidates together with symbols so that the candidates can be distinguished, and a voice for recognizing a symbol that is displayed in parallel with the candidates to identify the displayed candidate A recognition device is further provided, and further, (2) a symbol corresponding to the scroll function is determined at the time of displaying the candidates, and when the symbol is uttered, the following symbols are placed in the order of certainty of probability. , Or (3) a symbol corresponding to a correction function is determined in the case where the voice recognition device erroneously recognizes at the time of confirming the candidate, and the symbol is displayed when the symbol is uttered. Is obtained is characterized in that return to the state of the selection. Hereinafter, it demonstrates based on the Example of this invention.

FIG. 1 is a block diagram for explaining an embodiment of a voice recognition device according to the present invention. In the figure, 1 is a microphone, 2 is a monosyllabic recognition device, 3 is a language processing device, and 4 is a control device. 5 is a display device. First, the monosyllabic recognition device 2 takes in a voice from the microphone 1 and outputs a plurality of candidates for one monosyllabic. Next, the language processing device 3 attaches a score indicating the certainty from the recognition result of the monosyllabic recognition device 2 and outputs a plurality of phrase candidates. Next, the display device 5 displays the plurality of phrases side by side in the order of scores with symbols for distinguishing the phrases. Then, under the control of the control device 4, the monosyllabic recognition device 2 recognizes that the user utters a symbol corresponding to a syllable, and specifies a correct syllable.

Next, the process of inputting a phrase using this apparatus will be described based on a specific example. For example, when "Mechatronics" is uttered into the microphone 1, the monosyllabic recognition device 2 generates the syllable lattice of Table 1.

[0007]

[Table 1]

In Table 1, for example, the first candidate according to the recognition rate when "M" is uttered is "me", the second candidate is "ne", and the third candidate is "n". Then, the linguistic processor 3 develops the syllables in the syllable lattice, evaluates each syllable from the recognition rate and the linguistic validity, and outputs the score with certainty. The output example is shown in Table 2.

[0009]

[Table 2]

Next, the control device 4 displays the score-added phrase candidate group on the display device 5. The display example is shown in Table 3. The candidates are displayed from the top in the order of high score, and "a", "i" ... are symbols at the left end to distinguish the candidates.
"E" is displayed. Further, the control device 4 enables the monosyllabic recognition device 2 to input a voice, and in this case, "i" corresponds to a correct candidate, and when uttering "i", the recognition device 2 recognizes it and the candidate is found. Determine.

[0011]

[Table 3]

The embodiment corresponding to claim 2 is the embodiment described above, and when there is no correct candidate when displaying the phrase candidate group, "su" is determined as the symbol corresponding to the scroll function. When "su" is uttered, it is recognized by the voice recognition device 2 and the phrase group having the next highest score is displayed.

Further, the embodiment corresponding to claim 3 is the above-mentioned embodiment, in which the recognition device 2 erroneously recognizes as "a" even though uttering "i" corresponding to the correct candidate when determining the phrase candidate. If the finalized candidate is different, “te” is predetermined as a symbol corresponding to the correction function, and when the “te” is uttered, the voice recognition device 2 recognizes it and displays the phrase candidate group to display the candidate. It returns to the state of choosing.

[0014]

As is apparent from the above description, the present invention has the following effects. (1) In the voice recognition device according to the first aspect, since the phrase candidate can be selected by voice, the phrase can be input by a simpler operation. (2) In the voice recognition device according to claim 2, since the next phrase candidate group can be displayed and selected by voice, it is possible to select from a wider range of candidates. (3) In the voice recognition device according to the third aspect, when the candidate is confirmed, the candidate once confirmed by voice can be released, so that a correct candidate can be input even if the recognition device makes a mistake when confirming.

[Brief description of drawings]

FIG. 1 is a block diagram for explaining an embodiment of a voice recognition device according to the present invention.

[Explanation of symbols]

1 ... microphone, 2 ... monosyllabic recognition device, 3 ... language processing device,
4 ... Control device, 5 ... Display device.

Claims (3)

[Claims] 1. A monosyllabic recognition device which outputs a plurality of recognition results for one monosyllabic, and a language processing device which performs a language process on the recognition result and outputs a score indicating a certainty for each syllable. A display device that displays the phrase candidates together with the symbols so that the candidates can be distinguished in the order of certainty score, and if a symbol displayed in parallel with the candidate is uttered to identify the displayed candidate, this is displayed. A voice recognition device comprising a voice recognition device for recognition. 2. The voice recognition device according to claim 1, wherein
A voice recognition device characterized in that when a candidate is displayed, a symbol corresponding to a scroll function is determined, and when the symbol is uttered, the next group of candidates is displayed in the order of certainty score. 3. The voice recognition device according to claim 1, wherein
A voice recognition device characterized in that, when a voice recognition device makes an erroneous recognition when a candidate is determined, a symbol corresponding to a correction function is determined, and when the voice is uttered, the candidate display / selection state is restored.