A novel approach to vocoders: Scaled speech coder (SSC)

A novel approach to speech coding is introduced in this paper for the coding of speech signals used in vocoders. While conventional methods code the original input speech signal, the proposed algorithm codes the timescale modified signal instead. In this method, the transmitter speeds up original input speech signal by a factor of beta using waveform similarity overlap-addmethod (WSOLA). This signal is then coded using any of the low-bit rate speech coding algorithms. At the receiver side, the coded signal is first decoded to obtain the timescale modified signal. This signal is then slowed by a factor of 1/beta in order to reconstruct the original signal. The advantage stems from the fact that the time-scale modified signal requires half the coding time (when beta=0.5) than that of the original input speech. Before the realisation of the proposed system, the performances of three different time-scale modification (TSM) algorithms were evaluated through a series of subjective listening tests. Furthermore, it has been demonstrated that the time-scale modification of speech signals, before coding, reduces as much as half of the bit rate over conventional methods and produces high quality speech.