| 连续语流中声学地标的检测与分割 |
| |
| 引用本文： | 刘晶.连续语流中声学地标的检测与分割[J].茂名学院学报,2014(3):41-43. |
| |
| 作者姓名： | 刘晶 |
| |
| 作者单位： | 广东石油化工学院计算机与电子信息学院,广东茂名525000 |
| |
| 摘 要： | 在语音识别系统中，针对汉语普通话语音特点，增加采用分层级多参数加权综合的检测方法，以感知线索为依据，对连续语音流中的辅音性、阻塞性、擦音性、送气性、响音性、延续性、鼻音性、元音性、后位性等区别特征和语音特征的声学地标进行检测和分割。算法充分考虑了不同说话人、语音语境、语速和说话风格对声学地标的影响，这有助于提高检测和分割的准确率和鲁棒性。
|
| 关 键 词： | 声学地标 区别特征 语音识别 连续语流 |
Parallel Computing Based on the PVM |
| |
| LIU Jing.Parallel Computing Based on the PVM[J].Journal of Maoming College,2014(3):41-43. |
| |
| Authors: | LIU Jing |
| |
| Institution: | LIU Jing (College of Computer and Electronic Information, Guangdong University of Petrochemical Technology, Maoming 52.5000, China) |
| |
| Abstract: | According to mandarin phonetic features in speech recognition system, the hierarchical level, multi- parameter and weighted integration detection method can be additionally used in the system. Based on sensory clues, the acoustic landmarks of distinctive features and phonetic characteristics in the continuous speech flow of consonants, obstruction, frication, aspiration, sound, continuity, nasal, vowels, and retro- position are tested and divided. Algorithm fully comiders the different speakers, phonetic context and speed, speaking style influence on acoustic landmarks, which helps to improve the detection and segmentation accuracy and robustness. |
| |
| Keywords: | Acoustic landmarks Distinctive features Speech recognition Continuous speech flow |
| 本文献已被 维普 等数据库收录！ |
|
