Improved multiple descriptions sinusoidal speech coder adaptive to packet-loss rate

Multiple Descriptions (MD) speech coder can provide robust speech communication over unreliable channels such as the internet. However, existing MD speech coders use fixed scheme for different packet-loss environments. A novel Adaptive Multiple Descriptions Sinusoidal Coder (AMDSC) is proposed, which can vary in coding multiple descriptions according to the network packet-loss rate with optimally adding redundancy between two descriptions in order to make the final distortion minimum. Simulation results show that the proposed AMDSC outperforms existing MD speech coders by taking network loss characteristics into account.