الفهرس | Only 14 pages are availabe for public view |
Abstract Recent years have witnessed significant progress in the compression of speech. This has been the result of an inter working of coding theory and signal processing. Advances in compression have been accompanied by numerous standards for digital encoding and communication. The generic problem in signal compression is to minimize the bit rate in the digital representation of signal. while maintaining required levels of signal quality. This thesis consist of two parts, the first part is devoted to study of a multitude speech coding methods from point of view of time and spectral waveform, and study the bit rate compression against the speech quality and coding delay for each of them. We put in mind the complexity of each method. In the second part of this thesis, the study of new generation of bit reduction method which is called code excitation linear prediction CELP coder, we implemented the coder with bit rate 4.8 kb/s by using and testing the prediction technique. and codebook design. A low delay version of the coder called lattice low delay CELP was tested and implemented. Also, a modification in the coder design to reduce the computation complexity is proposed in this thesis. Perfonnance comparison between these two different cases \vas carried out and evaluated objectively and subjectively by using real speech signal. |