Revisiting IPA-based Cross-lingual Text-to-speech
This work addresses cross-lingual voice cloning for speech synthesis, but it is incremental as it revisits and empirically tests existing IPA-based approaches.
The paper tackles the problem of cross-lingual voice cloning using International Phonetic Alphabet (IPA) inputs in text-to-speech systems, finding that IPA processing methods have negligible impact, but using one speaker per language can fail due to information leakage.
International Phonetic Alphabet (IPA) has been widely used in cross-lingual text-to-speech (TTS) to achieve cross-lingual voice cloning (CL VC). However, IPA itself has been understudied in cross-lingual TTS. In this paper, we report some empirical findings of building a cross-lingual TTS model using IPA as inputs. Experiments show that the way to process the IPA and suprasegmental sequence has a negligible impact on the CL VC performance. Furthermore, we find that using a dataset including one speaker per language to build an IPA-based TTS system would fail CL VC since the language-unique IPA and tone/stress symbols could leak the speaker information. In addition, we experiment with different combinations of speakers in the training dataset to further investigate the effect of the number of speakers on the CL VC performance.