Two types of speech compression systems using inverse filter codebook
Yasuo Matsuyama · Electronics and Communications in Japan (Part I Communications) · 1981
Abstract Two types of speech compression systems which use a set of precalculated inverse filters (the inverse filter codebook) are studied. The information theoretic ideas of universal coding and block quantization motivated the use of the inverse filter code‐book. The first system finds the indices of the best matching inverse filter, the gain and the pitch interval for each frame of the input speech signal. In this case, a system operating at 800 bit/s for speech sampled at 8 kHz is adopted as the basic system. The second system is a waveform encoder which uses the inverse filter code‐book and tree searching. The system uses a single tree search, which is an improved version of the parallel tree search presented by the author, and a system of 8,875 bit/s is adopted as the basic device. Application of the inverse filter codebook is effective in achieving rate reduction and/or decreasing the system complexity. The first system falls into the class of vocoders and the second is a waveform encoder. The data rates of the two differ considerably, yet both types of systems are necessary. An example explaining this fact is also given in the text.