The Investigation of Vowel Recognition for the Optimal Mixed Cepstrum Feature
碩士 === 中興大學 === 統計學研究所 === 99 === This paper is mainly to recognize 337 isolated mandarin words for speaker-dependent. We use the linear predictive cepstrum feature and Mel-Frequency cepstrum feature for each part of consonant and vowel. We would like to find the optimal combination for linear...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Published: |
2011
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/47384103463915434812 |
| Summary: | 碩士 === 中興大學 === 統計學研究所 === 99 === This paper is mainly to recognize 337 isolated mandarin words for speaker-dependent. We use the linear predictive cepstrum feature and Mel-Frequency cepstrum feature for each part of consonant and vowel. We would like to find the optimal combination for linear predictive cepstrum feature and Mel-frequency cepstrum feature base on the recognition rate.
Furthermore, we also investigate the recognition rate for the change of duration on the time of consonant. In all the cases, the highest recognition rate in this paper is 97.44%.
|
|---|