Development of pronunciation training method using speech recognition for the hearing impaired

点击翻译按钮获取中文摘要

We added a vowel region estimation function to the existing consonant region estimation function in the VoiceAnalyzer speech analysis system, which made it easier to understand the configuration of phonemes in words.We further facilitated understanding by adding a hiragana display to the alphabetical display of the phonetic segment labels. Moreover, usability was enhanced through compatibility with the touch-input function newly added in Windows 8. We collected data of 10 persons × 235 words spoken by hearing-impaired people and found that the utterance of voiced speech is generally difficult for hearing-impaired people. We also understood thatspeech omitting phonemes or vague speech was improved through this system’ s visualization of sound.