ABSTRACT
Evaluating voice coder quality is a difficult task. No simple formula or mathematical calculation can provide a score indicating the quality of the decoded output speech. The problem lies in the fact that speech quality is inherently tied to speech perception. The perceived quality and understandability of coded speech depends on numerous conditions including speech content, speaker individuality, background noise, coding channel losses, and the listener. Removing some of these variables and averaging over the others lessens the scope of the discrepancies; however, different listeners will disagree on the quality of a single example of decoded speech.
