Design of voice compression algorithms and implementation of a voice over IP application
LE3 .A278 2007
Bachelor of Computer Science
This thesis presents a Voice over Internet Protocol (VoIP) system. The purpose of this system is to enable high-speed Internet users to have real-time conversation with each other, as smoothly and clearly as possible. Towards this end, this thesis proposes two novel approaches to compress human voice and analyzes the pros and cons of each method. One voice compression approach is a prediction-based method which uses known voice data to build up other data, then uses this new data for further prediction. Another method borrows the Human coding idea, which exploits the non-uniformity of the distribution of the difference between two adjacent samples in the voice data. In addition, this thesis compares the novel approaches to other open-source codecs by calculating numerical measurements of the distortion generated by the codecs. Finally, the design and implementation of a VoIP application are discussed. The aim of the design is to ensure that the system could be easily integrated with any of the two compression methods as well as others. The design has a technique to minimize lag that may occur in a real-time conversation. In addition, the design has a mechanism to analyze the output of different codecs every time a data frame is transferred. Based on this analysis, the best codec (in terms of compression ratio and audio quality) is chosen to compress that particular data frame. The VoIP application has been implemented in C for the Linux environment following the design. The application works well with both voice compression methods to provide a real-time conversation in a local area environment, which demonstrates the efficiency of the two compression algorithms and the overall system design.
The author grants permission to the University Librarian at Acadia University to reproduce, loan or distribute copies of my thesis in microform, paper or electronic formats on a non-profit basis. The author retains the copyright of the thesis.