LPC Synthesis from speech inputs containing: quantizing noise and additive white noise
M. R. Sambur, Nikil S. Jayant · The Journal of the Acoustical Society of America · 1975
An important problem in some communication systems is the performance of Linear Prediction analysis with speech inputs that have been corrupted by (signal-correlated) quantization distortion or additive white noise. To gain a first insight into this problem, a high-quality speech sample was deliberately degraded by using various degrees of adaptive differential PCM (ADPCM coding (two or more bits of quantization per sample) and by the introduction of additive white noise. The resulting speech samples were then analyzed to obtain the LPC control signals: pitch, gain,and the linear prediction coefficients. A distance metric proposed by Itakura was used to compare the original LP coefficients with the coefficients measured from the degraded speech. Other statistical techniques were used to compare the various measurements of pitch. In addition, the measured control signals were used to synthesize speech for perceptual evaluation, Results suggest that LPC analysis/synthesis is fairly immune to the degragation of ADPCM quantization. The effects of additive white noise are, however, more severe.