Effects of Smoothing and Quantizing the Parameters of Formant-Coded Voiced Speech

Abstract

A subjective evaluation of the effects of smoothing and quantization of the parameters for formant-coded voiced speech has been carried out. Thresholds were obtained for the degree of smoothing or quantization of specified parameters associated with a just noticeable degradation of all-voiced sentence material. Average quantization threshold levels obtained were: pitch period, 3.9 bits; first formant, 2.7 bits; second formant, 3.1 bits. The average of low-pass filter bandwidth thresholds for the simultaneous smoothing of formants was 7.6 Hz per formant; for pitch-period data the low-pass filter bandwidth threshold was 3.9 Hz. With simultaneous low-pass filtering and quantization, the estimated over-all bit rate for the coded parameters of all-voiced speech was approximately 200 bits/sec.