Abstract
This paper examines the accuracy rate of voice-input software, focusing particularly on the nature and causes of tone recognition errors. Our test results show that tone recognition errors account for 19.4% of all words tested. Most of these errors are in recognizing atonal and unstressed syllables. Our statistical analysis reveals that three main factors lead to errors in tone recognition: (1) different tonal types, (2) different dialectal backgrounds, and (3) different registers. This paper provides references for improvement of voice-input software.
| Keywords: | speech recognition; voice-input; tone; statistical analysis |
|---|
[Chinese Version | Index | Applied Linguistics (Yuyan Wenzi Yingyong) | Other Journals | Subscription form | Enquiry ]
Mail any comments and suggestions to
hkier-journal@cuhk.edu.hk .