Prosody-controllable spontaneous TTS with neural HMMs
This work addresses the problem of generating natural-sounding spontaneous speech for TTS applications, offering incremental improvements by adding prosody control to an existing neural HMM system.
The paper tackled the challenge of modeling spontaneous speech in text-to-speech (TTS) systems, which is problematic due to reduced articulation and disfluencies, by proposing a prosody-controllable neural HMM-based TTS architecture that can learn from small datasets and reproduce expressive phenomena, with perceptual tests showing no degradation in synthesis quality.
Spontaneous speech has many affective and pragmatic functions that are interesting and challenging to model in TTS. However, the presence of reduced articulation, fillers, repetitions, and other disfluencies in spontaneous speech make the text and acoustics less aligned than in read speech, which is problematic for attention-based TTS. We propose a TTS architecture that can rapidly learn to speak from small and irregular datasets, while also reproducing the diversity of expressive phenomena present in spontaneous speech. Specifically, we add utterance-level prosody control to an existing neural HMM-based TTS system which is capable of stable, monotonic alignments for spontaneous speech. We objectively evaluate control accuracy and perform perceptual tests that demonstrate that prosody control does not degrade synthesis quality. To exemplify the power of combining prosody control and ecologically valid data for reproducing intricate spontaneous speech phenomena, we evaluate the system's capability of synthesizing two types of creaky voice. Audio samples are available at https://www.speech.kth.se/tts-demos/prosodic-hmm/