A 2.4 KBPS speech coding method based on fuzzy vector quantization
Yoshiaki Asakawa, Koji Yamasaki, Akira Ichikawa, K. Ozawa · International Conference on Acoustics, Speech, and Signal Processing · 2002
A low-bit-rate, high-quality speech coding method is proposed. It is based on a PSE (power-spectrum envelope) analysis-synthesis method that provides high-quality synthetic speech. Its parameters are quantized using complementary fuzzy vector quantization (C-FVQ), which is newly developed to reduce the bit rate with small quantization distortion. With C-FVQ, an input vector is represented as a linear combination of a small number of complementary codewords that minimize quantization distortion with optimal coefficients. Speech coded at 2.4 kb.p.s. using the most practical type of C-FVQ is of high quality.>