To provide a speech recognition device which has an improved recognition rate of speech recognition by optimizing a dictionary for word recognition and consequently improving the performance of speech recognition.
The speech recognition device which performs word recognition for an input speech from a user by using a dictionary is characterized in that: word recognition and phoneme recognition are carried out for the input speech and when the word recognition result and phoneme recognition result which are thus obtained are discrepant and the word recognition result is a correct answer, the phoneme recognition result is recognized as a synonym for the word recognition result on condition that the relation between the similarity of the phoneme recognition result to a word corresponding to a word recognition result as a correct answer and the similarity to other words meets prescribed conditions.
COPYRIGHT: (C)2005,JPO&NCIPI
WO/2018/051841 | MODEL LEARNING DEVICE, METHOD THEREFOR, AND PROGRAM |
JP2008504607 | Expandable voice commands |
JP7161361 | Information management system |
Takeshi Hashimoto
Masaki Ashizawa
JP2050198A | ||||
JP2028700A |
Kushibuchi Kazue
Next Patent: DICHLORO CYANO COMPOUND, LIQUID CRYSTAL COMPOSITION AND LIQUID CRYSTAL ELECTROOPTICAL ELEMENT