This paper presents a voice signal synthesis and coding method in MATLAB. A voice signal is acquisitioned via microphone and saved in a text file. The sound samples from the file are coded with LPC (Linear Predictive Coding), LSF (Line Spectral Frequencies) and DFT (Discrete Fourier Transform) methods. This voice signal synthesis and coding is very important in today's communication via mobile phones. After the all operations the voice signal can be reproduced.

Human Voice Signal Synthesis and Coding

In order to decrease the LP spectral degradation in the USFS1016 4.8 kb/s CELP speech coder, the application of robust sample-selective LP parameter estimation procedures based on a quadratic classifier of nonstationary signals is proposed. Additionally, applications of the quadratic classifier with optimal and heuristically decision thresholds are considered. Comparative experimental analysis is done based on the cepstral distance as an objective spectral measure. The obtained results show that the proposed robust procedures achieve a smaller LP spectral degradation in the standard CELP coder, compared to standard LP methods, sample-selective methods, and robust method based on Huber's (1994) M-estimation theory.

Robust sample-selective LP parameter estimation in USFS1016 coder based on quadratic classifier

The influence of nonuniform scalar quantization and the specific interpolation procedure of LSP parameters, as proposed in USA FED STD 1016 CELP 4800 b/s speech coder, to the LPC spectra of speech obtained by using four standard LPC methods: autocorrelation with Hamming window, covariance, modified covariance and lattice, is considered. Comparative experimental analysis is done referring to the results of three different spectral measures related to the RMS LOG spectral measure: likelihood ratio, cosh measure and cepstral distance. The experimental analysis shows that, in the sense of the global LPC spectral representation, the best results are obtained by using the modified covariance method. In this way, the orientation to the autocorrelation method with Hamming window, proposed in USA FED STD 1016 CELP coder, is not justified.

M. Obradovic

Papers

Towards optimal LPC spectral parameters estimation in standard CELP 4800 b/s speech coder