Effect of choice of probability distribution, randomness, and search methods for alignment modeling in sequence-to-sequence text-to-speech synthesis using hard alignment
This work addresses alignment modeling issues in TTS for speech synthesis applications, but it is incremental as it builds on existing hard-attention methods.
The paper tackled the problem of sampling methods for discrete alignment in hard-attention-based sequence-to-sequence text-to-speech synthesis, finding that deterministic search is preferred over stochastic search and the binary Concrete distribution improves robustness in stochastic search.
Sequence-to-sequence text-to-speech (TTS) is dominated by soft-attention-based methods. Recently, hard-attention-based methods have been proposed to prevent fatal alignment errors, but their sampling method of discrete alignment is poorly investigated. This research investigates various combinations of sampling methods and probability distributions for alignment transition modeling in a hard-alignment-based sequence-to-sequence TTS method called SSNT-TTS. We clarify the common sampling methods of discrete variables including greedy search, beam search, and random sampling from a Bernoulli distribution in a more general way. Furthermore, we introduce the binary Concrete distribution to model discrete variables more properly. The results of a listening test shows that deterministic search is more preferable than stochastic search, and the binary Concrete distribution is robust with stochastic search for natural alignment transition.