Davinci Processor-based Voice Algorithm Design And Optimization

Speech communication is the most important way of communication. In the system of communication, speech compressing algorithms is used to save the network bandwidth and memory space. ITU-T Recommendation G.723.1 is one of the speech compressing algorithms. G.723.1 supports two bit rates 5.3kbit/s and 6.3kbit/s, and is possible to switch between the two rates at any frame boundary. When in the application of speech communication, another problem comes up, echo of speech seriously undermines the quality of the speech. Echo canceling becomes another topic of this dissertation. In order to implement the G.723.1 speech codec and echo canceller, we adopt the Davinci^TMTMS320DM6446 as the hardware platform. TMS320DM6446 is one of the most powerful fixed-point DSP of TI and it is designed for the digital multimedia applications.This dissertation is based on the Davinci^TMplatform, and the main aim of dissertation is to implement and optimize the speech algorithms on the platform. The structure of this dissertation is described as below:Chapter 1: Generally describe the speech compressing algorithm and echo canceling algorithm.Chapter 2: Introduce the Davinci^TMTMS320DM6446 platform, focus on the architecture of TMS320DM6446 and the instruction set of C64x.Chapter 3: Analyze and design the ITU-T G.723.1 speech algorithm.Chapter 4: Analyze and design the echo cancelling algorithm, and compare the performance of different echo algorithmsChapter 5: Implement and optimize the algorithm on the Davinci^TMplatform from aspects of code level, algorithm level and assembly level etc.Chapter 6: Summarize the work of the dissertation.