In this paper, a new adaptive Fourier decomposition

\r\n(AFD) based time-frequency speech analysis approach is proposed.

\r\nGiven the fact that the fundamental frequency of speech signals often

\r\nundergo fluctuation, the classical short-time Fourier transform (STFT)

\r\nbased spectrogram analysis suffers from the difficulty of window size

\r\nselection. AFD is a newly developed signal decomposition theory. It is

\r\ndesigned to deal with time-varying non-stationary signals. Its

\r\noutstanding characteristic is to provide instantaneous frequency for

\r\neach decomposed component, so the time-frequency analysis becomes

\r\neasier. Experiments are conducted based on the sample sentence in

\r\nTIMIT Acoustic-Phonetic Continuous Speech Corpus. The results

\r\nshow that the AFD based time-frequency distribution outperforms the

\r\nSTFT based one.<\/p>\r\n",
"references": null,
"publisher": "World Academy of Science, Engineering and Technology",
"index": "International Science Index 79, 2013"
}