Author : Nikita Atul Malhotra and Nikunj Tahilramani

This Paper basically focuses on the analysis and synthesis of the speech signal which takes reference from the speech coding technique of Conjugate-Structure Algebraic Code Excited Linear Prediction (CS-ACELP) and also the concept of vector quantization. The proposed technique analyses the variation in weighted speech by the changes in the weight factor and also by implementing it with and without vector quantization. This algorithm is based on CELP coding technique which works on analysis by synthesis principle with the frame size of 80 samples (10 ms). Steganography is implemented in the LSB of the input cover speech which contributes to the high frequency component of the cover speech. The weighted speech signal is generated using both quantized and unquantized LP parameters, which is being tested on various grounds to basically observe and analyze its overall performance. At the end an analytic comparison between the original and the weighted speech signal both subjectively as well as objectively is done. The values of the two weight factors produce a considerable change in the weighted speech quality.

Keywords: CS-ACELP, quantized LP parameters, steganography, unquantized LP parameters, vector quantization

Article published in International Journal of Current  Engineering  and Technology, Vol.4,No.3 (June- 2014)