语音信号处理-2.pptVIP

Introduction to Speech Signal Processing Dr. Zhang Sen zhangsen@ Chinese Academy of Sciences Beijing, China * View speech signal in math. Can be described by continuous function, but Hard to find explicit analytical form Non-linear Non-stationary, time-varying Some parts like noise Some parts like pseudo-periodic signal View speech signal in physics Wave generated by vibration Transmitted in air/media Analysis approaches Divide-and-conquer Approximation and simplicity Transformation (TD-FD) Analysis purpose To find speech features Which are important, which are trivial Correlation between features How features change? How to to change original signal Features can be classified as Time-domain features Frequency-domain features Or Short-term features Long-term features Feature representation Numerical: Vector or Distribution Diagram: curve or image Windowing (frame) In short-term, non-stationary-stationary and Non-linear-linear (10ms-25ms) Window types Window shapes A few words on Window function Commonly used speech features Zero-crossing-rate (ZCR) Peaks Power and energy Correlation, auto-correlation, AMDF Formant Pitch Frequency spectrum Cepstrum and MFCC Linear Predictive Coefficients (LPC), LPCC ZCR Level-crossing-rate Peaks Power and energy Correlation, auto-correlation, AMDF To measure the similarity of two signals or to detect the periodicity of a signal Sum x(k+i)*x(k+m+i) in a range, where k is the reference point and m is the lags Center-clipping technique Auto-correlation peaks Auto-correlation show Formant LPC-FFT Formant displays Some typical formant values Pitch, fundamental frequency Referred to as F0, determine tone and prosody Pitch estimation methods Auto-correlation and AMDF Cepstrum LPC Peak detection Pitch smoothing methods Dynamic programming N-point smoothing filter HMM Pitch show The pitch of a3 by auto-correlation method Spectrogram Representation of a signal highlighting several of its properties based on short-time Fourier analysi