Phonetically Switched Tree coding of speech with a G.727 Code Generator

A simple phonetic classification method based on frame error energy level in combination with the AMR-VADI Voice Activity Detection algorithm is used to classify input speech into five different modes: Voiced, Onset, Unvoiced, Hangover and Silence. Each mode is coded at a suitable bit-rate using a Tree Coder with perceptual error weighting criteria and a G.727 Code Generator. Apart from efficient coding, the Tree coder smoothes transitions between different coding modes. At an average bit-rate just greater than 16 kbps the phonetically switched Tree coder produces speech quality equivalent to the G.727 ADPCM coder at 32 kbps for nearly a 50% bit-rate saving. Apart from bit-rate savings; bit-rate scalability, moderate delay, and good tandeming performance are also achieved.