Metraipolis › archive › dc_2b094e0d9151
Distill Locally, Schedule Globally: Flow Maps for Few-Step Text-to-Speech — arXiv:2609.36324
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.36324
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T17:32:33.154Z
- SHA-256 of the archived text
c86c5e84cd278d5b0da88d955cce43c822c9fee7936e70f989236e60e5e6a31b- Stable URL of this page
https://metraipolis.com/documento/dc_2b094e0d9151
How to cite
Distill Locally, Schedule Globally: Flow Maps for Few-Step Text-to-Speech — arXiv:2609.36324 — arXiv (en), https://arxiv.org/abs/2609.36324 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_2b094e0d9151 (Metraipolis archive dc_2b094e0d9151, sha256 c86c5e84cd278d5b0da88d955cce43c822c9fee7936e70f989236e60e5e6a31b)
BibTeX
@misc{dc2b094e0d9151,
title = {Distill Locally, Schedule Globally: Flow Maps for Few-Step Text-to-Speech — arXiv:2609.36324},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.36324},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_2b094e0d9151; sha256 c86c5e84cd278d5b0da88d955cce43c822c9fee7936e70f989236e60e5e6a31b},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.36324. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Flow-matching text-to-speech (TTS) models achieve high synthesis quality but require many neural function evaluations (NFEs) to integrate their generative trajectories. Recent few-step flow-map distillation approaches for TTS construct targets from numerically integrated teacher trajectories, creating a trade-off between target accuracy and training cost. We propose Local Flow-Map Distillation (LFMD), which adapts Eulerian Map Distillation to conditional TTS and avoids teacher trajectory integration during target construction. For inference, we derive a sampling schedule (TD-DP) from teacher dynamics and consistency of the learned maps, with a single cost graph supporting multiple NFE budgets without external audio-metric evaluation. Because scheduling offers no flexibility at one NFE, we refine this regime with alignment-aware temporal self-distillation using soft-DTW. Across Seed-TTS and LibriSpeech-PC, LFMD improves low-NFE synthesis over a matched integral-distillation baseline. On Seed-TTS, the refined student reaches 1.80% WER with 1-NFE, compared with 1.76% for its 32-NFE teacher. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-30 in cs.LG). Categorie arXiv: Sound (cs.SD) ; Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2609.36324 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-07T16:59:48.481Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-30?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).