Continuous speech recognition based on ICA and geometrical learning
文献类型:会议论文
作者 | Feng H (Feng Hao) ; Cao WM (Cao Wenming) ; Wang SJ (Wang Shoujue) |
出版日期 | 2006 |
会议名称 | 4th international conference on machine learning and cybernetics |
会议日期 | aug 18-21, 2005 |
会议地点 | guangzhou, peoples r china |
关键词 | MULTI-WEIGHTED NEURON |
页码 | 3930: 974-983 |
通讯作者 | feng, h, zhejiang univ technol, informat coll, inst intelligent informat syst, hangzhou 310032, peoples r china. 电子邮箱地址: zjhzfh@mail.zjxu.edu.cn |
中文摘要 | we investigate the use of independent component analysis (ica) for speech feature extraction in digits speech recognition systems. we observe that this may be true for recognition tasks based on geometrical learning with little training data. in contrast to image processing, phase information is not essential for digits speech recognition. we therefore propose a new scheme that shows how the phase sensitivity can be removed by using an analytical description of the ica-adapted basis functions. furthermore, since the basis functions are not shift invariant, we extend the method to include a frequency-based ica stage that removes redundant time shift information. the digits speech recognition results show promising accuracy. experiments show that the method based on ica and geometrical learning outperforms hmm in a different number of training samples. |
英文摘要 | we investigate the use of independent component analysis (ica) for speech feature extraction in digits speech recognition systems. we observe that this may be true for recognition tasks based on geometrical learning with little training data. in contrast to image processing, phase information is not essential for digits speech recognition. we therefore propose a new scheme that shows how the phase sensitivity can be removed by using an analytical description of the ica-adapted basis functions. furthermore, since the basis functions are not shift invariant, we extend the method to include a frequency-based ica stage that removes redundant time shift information. the digits speech recognition results show promising accuracy. experiments show that the method based on ica and geometrical learning outperforms hmm in a different number of training samples.; zhangdi于2010-03-29批量导入; zhangdi于2010-03-29批量导入; ieee systems, man & cybernet tcc.; hong kong polytechn univ.; hebei univ.; s china univ technol.; chongqing univ.; sun yatsen univ.; harbin inst technol.; int univ germany.; zhejiang univ technol, informat coll, inst intelligent informat syst, hangzhou 310032, peoples r china; chinese acad sci, inst semicond, beijing 100083, peoples r china |
收录类别 | 其他 |
会议主办者 | ieee systems, man & cybernet tcc.; hong kong polytechn univ.; hebei univ.; s china univ technol.; chongqing univ.; sun yatsen univ.; harbin inst technol.; int univ germany. |
会议录 | advances in machine learning and cybernetics丛书标题: lecture notes in artificial intelligence |
会议录出版者 | springer-verlag berlin ; heidelberger platz 3, d-14197 berlin, germany |
学科主题 | 人工智能 |
会议录出版地 | heidelberger platz 3, d-14197 berlin, germany |
语种 | 英语 |
ISSN号 | 0302-9743 |
ISBN号 | 3-540-33584-6 |
源URL | [http://ir.semi.ac.cn/handle/172111/9990] |
专题 | 半导体研究所_中国科学院半导体研究所(2009年前) |
推荐引用方式 GB/T 7714 | Feng H ,Cao WM ,Wang SJ . Continuous speech recognition based on ICA and geometrical learning[C]. 见:4th international conference on machine learning and cybernetics. guangzhou, peoples r china. aug 18-21, 2005. |
入库方式: OAI收割
来源:半导体研究所
浏览0
下载0
收藏0
其他版本
除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。