Advances in Spectral Parameterization for Statistical (HMM-Based) TTS

HMM-based parametric speech synthesis has recently become an alternative to the concatenative TTS approach, especially when low footprint and general speech domain are required. A majority of speech parameterization models used in state-ofthe art HMM TTS systems employ source-filter waveform synthesis schemes. Sinusoidal representation and waveform generation of speech is an alternative to the source-filter model, which is successfully applied in speech coding, unitselection TTS and voice conversion, but rarely used for HMM TTS systems. In this paper we utilize Regularized Cepstral Coefficients (RCC) estimated in mel-frequency scale for sinusoidal amplitude spectrum envelope modeling within an HMM-based TTS framework. Improved subjective quality for mel-frequency RCC (MRCC) combined with the sinusoidal model based reconstruction is reported and compared to the state-of-the-art MGC-LSP parameters.