New approach to speech digitization combining time domain harmonic scaling and adaptive residual coding
Arun Keshav Pande · 1981
A wide array of digital speech techniques exists today, requiring binary data rates varying from 2.4 kb/s to 64 kb/s. Particularly, there is a growing interest in developing the speech transmission systems that operate at intermediate data rates of 8 kb/s to 16 kb/s. This thesis presents a new approach to speech digitization at the bit rate of 9.6 kb/s using time domain harmonic scaling and adaptive residual coding. The objective of the coder design is to provide good speech quality and a robust performance for channel bit error rates up to 1%. A number of issues were investigated, including: system configuration; frequency scaling schemes; use of fixed word length code; transmission error studies; and buffer control strategy. The simulation studies have indicated that this system produces excellent speech quality. The system complexity is kept to a minimum so that it can be implemented in real-time using current technology.