| Tone features in whispered Chinese |
| |
| 引用本文： | LI Xueli,XU Boling. Tone features in whispered Chinese[J]. 自然科学进展(英文版), 2005, 15(3): 285-288 |
| |
| 作者姓名： | LI Xueli XU Boling |
| |
| 作者单位： | State Key Laboratory of Modem Acoustics, Institute of Acoustics, Nanjing University, Nanjing 210093, China;School of Information Science and Engineering, Shandong University, Jinan 250100, China,State Key Laboratory of Modem Acoustics, Institute of Acoustics, Nanjing University, Nanjing 210093, China |
| |
| 摘 要： | Study on the whispered tone is important to speech recognition and conversion in whispered Chinese. In this paper, the characteristics of whispered speech are introduced and the tone features in whispered Chinese are discussed. There is no fundamental frequency in the whispered speech, so other features, such as the amplitude envelope, duration, glottal area, lip area, forrnant, and vocal tract length, are extracted and their contributions to the automatic tone recognition are compared. From the experiments with six simple Chinese whispered vowels in four tones, it is proved that loudness-weighted 32 Mel-frequency bands log-amplitude envelopes and duration can be used as the main tone features in the whispered Chinese tone recognition. The average tone recognition rate approaches that of the human perception level.
|
| 关 键 词： | whispered speech tone feature loudness-weighted Mel-frequency band amplitude envelope sound dur |
Tone features in whispered Chinese |
| |
| LI Xueli,XU Boling. Tone features in whispered Chinese[J]. Progress in Natural Science, 2005, 15(3): 285-288 |
| |
| Authors: | LI Xueli XU Boling |
| |
| Affiliation: | 1. State Key Laboratory of Modem Acoustics, Institute of Acoustics, Nanjing University, Nanjing 210093, China;School of Information Science and Engineering, Shandong University, Jinan 250100, China
2. State Key Laboratory of Modem Acoustics, Institute of Acoustics, Nanjing University, Nanjing 210093, China |
| |
| Abstract: | Study on the whispered tone is important to speech recognition and conversion in whispered Chinese. In this paper, the characteristics of whispered speech are introduced and the tone features in whispered Chinese are discussed. There is no fundamental frequency in the whispered speech, so other features, such as the amplitude envelope, duration, glottal area, lip area, forrnant, and vocal tract length, are extracted and their contributions to the automatic tone recognition are compared. From the experiments with six simple Chinese whispered vowels in four tones, it is proved that loudness-weighted 32 Mel-frequency bands log-amplitude envelopes and duration can be used as the main tone features in the whispered Chinese tone recognition. The average tone recognition rate approaches that of the human perception level. |
| |
| Keywords: | whispered speech tone feature loudness-weighted Mel-frequency band amplitude envelope sound duration |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
| 点击此处可从《自然科学进展(英文版)》浏览原始摘要信息 |
|
点击此处可从《自然科学进展(英文版)》下载全文 |
|
