Phonological Features for 0-shot Multilingual Speech Synthesis
This addresses the challenge of multilingual speech synthesis for code-switching, which is prevalent globally, but is incremental as it modifies existing TTS models rather than introducing a new paradigm.
The paper tackled the problem of enabling code-switching in text-to-speech (TTS) for languages unseen during training by using phonological features derived from the International Phonetic Alphabet (IPA), resulting in the generation of intelligible, code-switched speech and approximation of unseen sounds.
Code-switching---the intra-utterance use of multiple languages---is prevalent across the world. Within text-to-speech (TTS), multilingual models have been found to enable code-switching. By modifying the linguistic input to sequence-to-sequence TTS, we show that code-switching is possible for languages unseen during training, even within monolingual models. We use a small set of phonological features derived from the International Phonetic Alphabet (IPA), such as vowel height and frontness, consonant place and manner. This allows the model topology to stay unchanged for different languages, and enables new, previously unseen feature combinations to be interpreted by the model. We show that this allows us to generate intelligible, code-switched speech in a new language at test time, including the approximation of sounds never seen in training.