A study on a new non-uniform speech coding using the components of separated by harmonics and formants frequencies

Abstract

In speech processing, coding may be define as a digital representation of speech sound that provides efficient storage, transmission, recovery, and perceptually faithful reconstruction of the original speech. In the coding methods, the waveform coding method can maintain the high quality and personality, but the major drawback to this method is large amount of data required. Therefore, we suggest a waveform coding method using the nonuniform sampling and separated bandwidth of speech signals. Results show that this algorithm is robust in noisy environments, and we can get the average compression ratios of 5.67 and the gains of 3dB in our algorithm for original speech. © 2012 Springer-Verlag.