#natural language process...
Jan 2026
SITA: Learning Speaker-Invariant and Tone-Aware Speech Representations for Low-Resource Tonal Languages
Against multilingual, speaker-adversarial, label-aware, and semi-supervised baselines, SITA achieves the best trade-off between cross-gender lexical retrieval and tone separation, while maintaining ASR accuracy close to an ASR-adapted XLS-R teacher.
Tianyi Xu, Xu Ouyang, Binwei Yao et al.
· arXiv.org · 2 citations
· ⚡1