Time-Domain Envelope Modulating the Noise Component of Excitation in a Continuous Residual-Based Vocoder for Statistical Parametric Speech Synthesis

Abstract

In this paper, we present an extension of a novel continuous residual-based vocoder for statistical parametric speech synthesis. Previous work has shown the advantages of adding envelope modulated noise to the voiced excitation, but this has not been investigated yet in the context of continuous vocoders, i.e. of which all parameters are continuous. The noise component is often not accurately modeled in modern vocoders (e.g. STRAIGHT). For more natural sounding speech synthesis, four time-domain envelopes (Amplitude, Hilbert, Triangular and True) are investigated and enhanced, and then applied to the noise component of the excitation in our continuous vocoder. The performance evaluation is based on the study of time envelopes. In an objective experiment, we investigated the Phase Distortion Deviation of vocoded samples. A MUSHRA type subjective listening test was also conducted comparing natural and vocoded speech samples. Both experiments have shown that the proposed framework using Hilbert and True envelopes provides high-quality vocoding while outperforming the two other envelopes.

Ultrasound-Based Silent Speech Interface Built on a Continuous Vocoder

“…Next, we removed the post-processing step in the estimation of the MVF parameter and thus improved the modelling of unvoiced sounds within our continuous vocoder [29]. Finally, we applied various time domain envelopes for advanced modeling of the noise excitation [30].…”

“…During the synthesis phase, voiced excitation is composed of residual excitation frames overlap-added pitch synchronously, depending on the continuous F0 [28,29,30]. After that, this voiced excitation is lowpass filtered frame by frame at the frequency given by the MVF parameter.…”

RNN-based speech synthesis using a continuous sinusoidal model

“…Previous studies have shown that human voice can be modelled effectively as a sum of sinusoids and has shown the capability of providing high-quality copy synthesis and prosodic modifications [29] [30] [31]. Therefore, in [32] we proposed a continuous sinusoidal model (CSM) that is applicable in statistical frameworks by keeping the number of our vocoder parameters unchanged [26]. Experimental results from objective and subjective evaluations have shown that the proposed vocoder gives state-of-the-art vocoders performance in analysis-synthesis while outperforming the previous work of our continuous F0 based source-filter vocoder.…”

“…By keeping the number of our previous source-filter vocoder parameters unchanged [26] and similarly to [29] [43], the synthesis algorithm implemented in this paper decomposes the speech frames into a lower-band voiced component sv(t) and an upper-band noise component sn(t) based on MVF values. We define these components here as…”

“…In order to avoid discontinuities at the frames boundaries, Overlap-add (OLA) technique is used to reconstruct the speech signal from their corresponding parameters estimated from our analysis model in [26]. If the current frame is voiced, the harmonic part can be expressed as:…”