Speech-recognition system, storage medium, and method of speech recognition

1. A speech recognition system that recognizes speech data, the speech recognition system comprising:

a speech recognition part that performs speech recognition of the speech data, and calculates a likelihood of the speech data with respect to a registered word that is pre-registered;

a reliability judgment part that performs reliability judgment on the speech recognition based on the likelihood; and

a judgment reference change processing part that changes a judgment reference for the reliability judgment, according to an utterance speed of the speech data.

2. The speech recognition system according to claim 1 , wherein

the reliability judgment part performs the reliability judgment for judging the reliability of the speech recognition based on a comparison result between a likelihood difference judgment threshold and a likelihood difference that is a difference in the likelihood among a plurality of the registered words obtained as a result of the speech recognition, and

the judgment reference change processing part changes the likelihood difference judgment threshold to be used for the reliability judgment to have a greater value, as the utterance speed becomes slower.

3. The speech recognition system according to claim 2 , wherein

the likelihood difference judgment threshold is set corresponding to an acoustic model of each of the registered words, and

the reliability judgment part use the likelihood difference judgment threshold set corresponding to an acoustic model of a first registered word whose likelihood obtained as a result of the speech recognition is at the first rank, thereby judging the reliability of the speech data being the first registered word.

4. The speech recognition system according to claim 3 , wherein the judgment reference change processing part judges the utterance speed of the speech data based on the recognition time in speech recognition of the speech data and the number of vowels in an acoustic model of the registered word.

5. The speech recognition system according to claim 4 , wherein the reliability judgment part obtains the likelihood difference between the first registered word and a second registered word whose likelihood is at the second rank, and judges the reliability of the speech data being the first registered word based on the result of comparison between the likelihood difference and the likelihood difference judgment threshold.

6. The speech recognition system according to claim 1 , wherein the judgment reference change processing part determines the utterance speed.

7. A non-transitory computer-readable medium storing a speech recognition program, the speech recognition program that renders a computer to function as:

a speech recognition part that performs speech recognition of speech data, and calculates a likelihood of the speech data with respect to a registered word that is pre-registered;

a reliability judgment part that performs reliability judgment on the speech recognition based on the likelihood; and

a judgment reference change processing part that changes a judgment reference for the reliability judgment, according to an utterance speed of the speech data.

8. A speech recognition method for performing speech recognition of speech data, and the method comprising:

a speech recognition step of performing the speech recognition and calculating a likelihood of the speech data with respect to a registered word that is pre-registered;

a judgment reference change processing step of changing a judgment reference for reliability judgment according to an utterance speed of the speech data; and

a reliability judgment step of performing reliability judgment on the speech recognition based on the likelihood.