mlsanityData to 14 Oct 2026

Ready to Speak: Aligning LLMs for TTS-Friendly Text Generation

arXiv 2609.01246 ↗First posted 1 Sep 2026cs.CLcited by 0

On the map: LLM alignment / RLHF / preference optimisation → Preference alignment for recommenders and RAG

Abstract (arXiv)

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→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.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Preference alignment for recommenders and RAG, one of 10 directions in LLM alignment / RLHF / preference optimisation. Builds on 20 of its 36 references that are in the map. First posted on arXiv 1 Sep 2026.

Builds on

20 of 36 references.

Closest ancestors · 2

Same direction, newest first

  1. 2025Koel-TTS: Enhancing LLM based Speech Generation with Preference Alignment and Classifier Free Guidance7 Feb
  2. 2024Preference Alignment Improves Language Model-Based TTS19 Sep

Other references · 18

  1. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 9,379
  2. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 8,065
  3. 2023LIMA: Less Is More for Alignmentcited by 1,100
  4. 2023Prometheus: Inducing Fine-grained Evaluation Capability in Language Modelscited by 575
  5. 2024Prometheus 2: An Open Source Language Model Specialized in Evaluating Other Language Modelscited by 446
  6. 2025Reinforcement Learning with Verifiable Rewards Implicitly Incentivizes Correct Reasoning in Base LLMscited by 289
  7. 2024Towards Controllable Speech Synthesis in the Era of Large Language Models: A Systematic Surveycited by 23
  8. 2025Advancing Zero-shot Text-to-Speech Intelligibility across Diverse Domains via Preference Alignmentcited by 18
  9. 2025Constrained Discrete Diffusioncited by 17
  10. 2025Shot-by-Shot: Film-Grammar-Aware Training-Free Audio Description Generationcited by 13
  11. 2023A Chat About Boring Problems: Studying GPT-based text normalizationcited by 10
  12. 2024ELITR-Bench: A Meeting Assistant Benchmark for Long-Context Language Modelscited by 9
6 more
  1. 2024Speechworthy Instruction-tuned Language Modelscited by 7
  2. 2025FaST: Feature-aware Sampling and Tuning for Personalized Preference Alignment with Limited Datacited by 6
  3. 2022CLID: Controlled-Length Image Descriptions with Limited Datacited by 4
  4. 2025PolyNorm: Few-Shot LLM-Based Text Normalization for Text-to-Speechcited by 3
  5. 2025Beyond One-Size-Fits-All Summarization: Customizing Summaries for Diverse Userscited by 2
  6. 2025CIE: Controlling Language Model Text Generations Using Continuous Signalscited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Data-efficient Targeted Token-level Preference Optimization for LLM-based Text-to-Speech
  2. 2025DiffPO: Diffusion-styled Preference Optimization for Efficient Inference-Time Alignment of Large Language Models
  3. 2025Data Selection for LLM Alignment Using Fine-Grained Preferences
  4. 2025GOAT-TTS: Expressive and Realistic Speech Generation via A Dual-Branch LLM
  5. 2025Leveraging Robust Optimization for LLM Alignment under Distribution Shifts
  6. 2025On Evaluating LLM Alignment by Evaluating LLMs as Judges
  7. 2025Not All Preferences are What You Need for Post-Training: Selective Alignment Strategy for Preference Optimization
  8. 2026CATTO: Balancing Preferences and Confidence in Language Models