유효 음성 추출을 위한 머징알고리즘에 관한 연구 = A Study on the Merging Algorithm to Extract Valid Speech-Sound

다국어 초록 (Multilingual Abstract)

A valid speech-sound block can be classified to provide important information for speech recognition. The classification of the speech-sound block comes from the MRA(Multi-Resolution Analysis) property of the DWT(Discrete Wavelet Transform) which is used to reduce the computational time for the pre-procession of speech recognition. The merging algorithm is proposed to extract valid speech-sounds upon consideration of its position and frequency range. The merging algorithm needs some numerical methods for an adaptive DWT implementation and shows an unvoiced/voiced classification and a denoising. Since the merging algorithm can decide the processing parameters only at the standpoint of voices and is independent of system noises, it is more useful to extract valid speech-sounds. The merging algorithm has the adaptive feature for arbitrary system noises and an excellent denoising SNR (Signal-to-noise ratio).