This paper presents a speech coder operating at a very low bit-rate using a model called "phase-equalized pitch waveform". The basic idea of the coder is to employ pitchwise extraction of the linear predictive residual signal, a pitch waveform, in voiced speech. The residual signal is processed with a phase-equalization filter to increase the efficiency of both the pitch waveform quantization and interpolation. Listening tests showed that efficient and high-quality coding is achieved at 2.0 kbits/s. The quality of the coder is equal to that of the DoD FS1016 standard CELP at 4.8 or 2.4 kbits/s MELP.
ASJC Scopus subject areas
- Modelling and Simulation
- Language and Linguistics
- Linguistics and Language
- Computer Vision and Pattern Recognition
- Computer Science Applications