Metraipolis › archive › dc_dd250e23ed18
Ready to Speak: Aligning LLMs for TTS-Friendly Text Generation — arXiv:2609.01246
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.01246
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-17T06:41:32.811Z
- SHA-256 of the archived text
98f066bbdaf93dd090ffba72d1a0e9700665f68d2b69c7cd37a9c20519f58c1c- Stable URL of this page
https://metraipolis.com/documento/dc_dd250e23ed18
How to cite
Ready to Speak: Aligning LLMs for TTS-Friendly Text Generation — arXiv:2609.01246 — arXiv (en), https://arxiv.org/abs/2609.01246 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_dd250e23ed18 (Metraipolis archive dc_dd250e23ed18, sha256 98f066bbdaf93dd090ffba72d1a0e9700665f68d2b69c7cd37a9c20519f58c1c)
BibTeX
@misc{dcdd250e23ed18,
title = {Ready to Speak: Aligning LLMs for TTS-Friendly Text Generation — arXiv:2609.01246},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.01246},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_dd250e23ed18; sha256 98f066bbdaf93dd090ffba72d1a0e9700665f68d2b69c7cd37a9c20519f58c1c},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.01246. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Current Large Language Models (LLMs) are primarily optimized for written text, often producing outputs that are grammatically correct and helpful yet poorly suited for spoken delivery via Text-to-Speech (TTS). In this work, we study how to make LLMs natively generate TTS-friendly text, which we frame as a preference alignment problem: instead of relying on downstream rewriting modules, we directly align LLMs to generate text optimized for spoken delivery. We introduce two preference datasets spanning different target domains, CORA and Recipe, which contain paired TTS-friendly and TTS-unfriendly responses. We further propose an evaluation suite combining a pattern-based heuristic metric, a TTS$\to$ASR evaluation pipeline, and a MUSHRA listening study with human judges. Our experiments compare the recently proposed Feature-aware Sampling and Tuning (FaST) framework -- leveraging interpretable features instead of a black-box reward model -- against an array of alignment baselines on the TTS-friendly generation task. Notably, we found that FaST achieves the best overall tradeoff between TTS-friendliness and helpfulness across various settings. We also identified a strong correlation between our different metrics, highlighting the ability to reliably assess TTS-friendliness via an efficient heuristic. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-02 in cs.CL). Categorie arXiv: Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2609.01246 — 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 2026-10-17T06:33:38.852Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-02?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).