Two-channel sound source localization method for speech enhancement system음질 개선 시스템을 위한 두 채널 음원의 방향 추적 방법

Cited 0 time in webofscience Cited 0 time in scopus
  • Hit : 581
  • Download : 0
Human can distinguish the desired speech signal in the environment of the signal and noise mixture. Many researchers have endeavored to analyze and implement this human`s hearing capability in signal processing fields, however still can`t get the satisfied result. With the developing of the network and the Information Technology, many researches for the extended speech communication and interface services such as the video telephony and speech recognition system are actively processed, however the service`s quality still remains as a problem. One of the major reasons is that the real environment with the background noise is quite different from the noiseless laboratory where systems can work properly. So the performance of the systems drops drastically in the real environment. The speech enhancement means that it can extract the desired signal from the one or more sources and the noise mixture. By using this technology as a pre-processor of the speech processing system such as the speech communication system or the speech recognition system, we can increase the system`s performance. The speech enhancement system is classified by two fields depending on the number of the input signals. The single channel speech enhancement system uses only one input signal and is very commonly commercialized. Multi-channel speech environment system uses multiple input signals. The single channel speech enhancement works poorly when it fails to estimate the environment`s noise information, and the multi-channel speech enhancement requires more hardware cost and computation capability to process the multiple input signals and works poorly when it fails to estimate the desired signal`s direction. Some of these problems can be resolved by detecting the voice activity or estimating the signal`s direction accurately. We propose two-step method of increasing the accuracy of direction-of-arrival (DOA) estimation by using two input signals to complement the problems of the curr...
Advisors
Yoon, Hyun-Sooresearcher윤현수researcher
Description
한국과학기술원 : 전산학전공,
Publisher
한국과학기술원
Issue Date
2008
Identifier
303643/325007  / 000995334
Language
eng
Description

학위논문(박사) - 한국과학기술원 : 전산학전공, 2008. 8., [ ix, 76 p. ]

Keywords

Sound Source Localization; Speech Enhancement; 음원 방향; 음질 개선; Sound Source Localization; Speech Enhancement; 음원 방향; 음질 개선

URI
http://hdl.handle.net/10203/33257
Link
http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=303643&flag=dissertation
Appears in Collection
CS-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