Reproduce “Towards Natural Bilingual and Code-Switched Speech Synthesis Based on Mix of Monolingual Recordings and Cross-Lingual Voice Conversion” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Shengkui Zhao, Trung Hieu Nguyen, Hao Wang, Bin Ma
The paper expands two monolingual recording sets into bilingual training data for English–Mandarin speech synthesis. Cross-lingual voice conversion renders each speaker’s recordings in the other language, and synthetic code-switched speech augments training for three TTS architectures. For the Mandarin target speaker, Transformer code-switched speech received a mean opinion score of 3.97 ± 0.08; subjective results cover only two speakers.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.