| title | Speaker-Disentangled Chunk-Wise Regression for Syllabic Tokenization |
|---|---|
| authors | Ryota Komatsu, Kota Kawakita, Takuma Okamoto, Takahiro Shinozaki |
| date | 2026-07-05 |
| venue | arXiv (preprint, not peer reviewed) |
| id | arXiv:2607.04064 |
| upvotes | 1 (Hugging Face Daily Papers, 2026-11-01) |
| links | abs pdf hf code |
| abstract | Unsupervised syllabic tokenization aims to learn discrete syllabic tokens that capture latent linguistic content-related structure from raw speech. Recent syllabic tokenization methods employ teacher-student distillation of the pretrained HuBERT to organize latent speech frame representations into syllabic segments. However, when trained with an utterance-level cross-entropy objective, the model predicts speaker identity rather than linguistic content, thereby compromising the purity of syllabic tokens. To address this problem, we propose a speaker-disentangled syllabic tokenizer that regresses speaker-perturbed student representations toward clean teacher targets within fixed-length chunks. Experimental results demonstrate that our proposed method achieves state-of-the-art performance in syllable boundary detection and syllabic segment clustering. Moreover, a speech language model trained on our syllabic tokens achieves a 7% relative improvement in syntactic and semantic understanding over the phone-level SpiRit-LM. |
cite
@misc{komatsu2026speaker,
title={Speaker-Disentangled Chunk-Wise Regression for Syllabic Tokenization},
author={Ryota Komatsu and Kota Kawakita and Takuma Okamoto and Takahiro Shinozaki},
year={2026},
eprint={2607.04064},
archivePrefix={arXiv},
url={https://arxiv.org/abs/2607.04064}
}