Study on cochlear model and speech recognition system based on stimulus frequency otoacoustic emission = 자극주파수 이음향방사에 기반한 와우각 모델 및 음성인식 시스템에 관한 연구

Cited 0 time in webofscience Cited 0 time in scopus
  • Hit : 361
  • Download : 0
Automatic speech recognition (ASR) system is one of good solutions in human-machine interface. As machines are getting more complex and harder to operate, easier and more natural operation is required, hence ASR system has been emphasized. However, the conventional ASR system does not have acceptable performance yet in real world. To make a better ASR performance, researchers applied various concepts and philosophies on the system. Among the approaches, one of the most powerful ways is to model human hearing system and apply it to ASR core part because none of existing ASR system ever reached the recognition rate of human hearing ability, especially in noisy environment. The essential part of human hearing is a cochlea of inner ear. If we model the cochlea activities and find out what plays the main role in the hearing, ASR performance can be enhanced. The matter is how we get the information of the cochlea activity non-invasively and efficiently. One way is to use OtoAcoustic Emission(OAE). Especially, stimulus frequency otoacoustic emissions (SFOAEs) are more proper to investigate cochlea conditions because only one frequency component is used for the generation of SFOAE and we may ignore interactions of several frequency components. The steps for building human-like ASR system may consist of (1) measurements of SFOAEs, (2) building SFOAE generation model with measured data, and (3) constructing feature extraction algorithms based on the cochlear characteristics. Thus, in this study, an efficient method for measuring SFOAEs was developed incorporating (1) stimulus with swept frequency or level and (2) the digital heterodyne analysis. SFOAEs were measured for 550 $\sim$ 1450 Hz and stimulus levels of 32 to 62 dB SPL in 8 normal human adults. The mean level, number of peaks, frequency spacing between peaks, phase change, and energy-weighted group delays of SFOAEs in frequency sweeping experiments and SFOAE input-output (I/O) functions in level sweeping ...
Advisors
Lee, Soo-Youngresearcher이수영researcher
Description
한국과학기술원 : 바이오및뇌공학과,
Publisher
한국과학기술원
Issue Date
2008
Identifier
303565/325007  / 020025307
Language
eng
Description

학위논문(박사) - 한국과학기술원 : 바이오및뇌공학과, 2008. 8., [ viii, 91 p. ]

Keywords

Stimulus Frequency OtoAcoustic Emission; Active Cochlear Model; Speech Recognition; Automatic Gain Control; 자극주파수 이음향방사; 능동 와우각 모델; 음성 인식; 자동 이득 제어; Stimulus Frequency OtoAcoustic Emission; Active Cochlear Model; Speech Recognition; Automatic Gain Control; 자극주파수 이음향방사; 능동 와우각 모델; 음성 인식; 자동 이득 제어

URI
http://hdl.handle.net/10203/27065
Link
http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=303565&flag=dissertation
Appears in Collection
BiS-Theses_Ph.D.(박사논문)
Files in This Item
There are no files associated with this item.

qr_code

  • mendeley

    citeulike


rss_1.0 rss_2.0 atom_1.0