Reproduce “STCTS: Generative Semantic Compression for Ultra-Low Bitrate Speech via Explicit Text-Prosody-Timbre Decomposition” 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
Siyu Wang, Haitao Li, Donglai Zhu
STCTS reconstructs speech at the receiver from transmitted text, sparse prosody cues, and a reusable speaker embedding, rather than sending the waveform. It combines speech recognition, prosody extraction, and text-to-speech, with speaker information cached for reuse. On LibriSpeech test-clean, its high-quality mode uses 79.6 ± 8.9 bps and has NISQA 4.255 ± 0.407, but higher WER than Opus and EnCodec.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.