An investigation of speech quality from a DESA synthesis system

The perceived quality of the reconstructed speech signals from a discrete energy separation algorithm (DESA) based analysis and synthesis system is investigated. It will focus on how the size of the analysis frame affects the output speech quality. Since the DESA based system uses subbands, the number and bandwidths of subbands will remain constant. The method of encoding the signal will also remain constant. The size of the analysis frame is important in this system since it affects the compression ratio. An evaluation of the output signal will be done using an informal opinion survey by test subjects listening to the recordings. This evaluation will be performed using a seven-point system.