A HARMONIC+NOISE CODER WITH IMPROVED TRANSIENT SPEECH PERFORMANCE

摘要

This paper presents a 2.4 kbps harmonic+noise coder with improved transient speech performance. The coder operates in either the steady mode or the transient mode. A predictive phase model is proposed for the efficient encoding of phase information in steady mode. In this model, the normalized frequency deviations are used to minimize the phase prediction errors. A polynomial is fitted to the set of normalized frequency deviations and the polynomial coefficients are vector quantized in closed-loop. For the transient speech, the voiced part is modeled by the low-frequency harmonic components while the unvoiced part is modeled by an all-pole model. The transitional behavior is represented by the onset time and growth rate of each harmonic component. An analysis-by-synthesis procedure is proposed to determine the onset times and growth rates of the harmonic components in the time domain. The proposed coder retains the transitional characteristics of speech signal. Subjective testing shows that the speech quality of the proposed coder is sufficient for communication purposes.