الفهرس | Only 14 pages are availabe for public view |
Abstract The aim of all speech coding is to reach the best quality with the least bandwidth. At bit rates above 4 kbps, speech-specific hybrid coders based on code excited linear prediction (CELP) can produce good quality speech. But at bit rates less than 4 kbps it becomes very difficult. CELP is the quite efficient closed loop analysis-by-synthesis method. In this thesis, (CELP) is implemented using Matlab. The thesis aims to find the minimum number of linear predictive coefficients that yields to good codec quality and hence the bit rate is reduced. Arabic words are used. The quality of the codec is influenced by the language used. More compression with language or accents other than English leads to less quality. This coding technique is analysed on the basis of subjective and objective tests. Intelligibility, Mean Opinion Score (MOS), Signal-to-Noise Ratio (SNR) and rate distortion are employed. |