AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Evaluating and Rewarding LALMs for Expressive Role-Play TTS via Mean Continuation Log-Probability

Yong Ren, Jingbei Li, Haiyang Sun, Yujie Chen, Cheng Yi, Yechang Huang, Hao Gu, Ye Bai, Xuerui Yang

ICML 2026 · Large language models › Language, safety and interpretability · approach: Speech and audio language models · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add SFT on WenetSpeech-RP-TTS to Step-Audio-2-mini (ablation)Step-Audio-2-mini · SubjectiveMOS: 1.856 -> 3.178Improved
add SFT (ablation)Step-Audio-2-mini · W/ audio historyMCLP: -4.829 -> -4.725Improved
add SFT (ablation)Step-Audio-2-mini · W/o audio historyCER (%): 12.007 -> 6.306Improved
add SFT (ablation)Step-Audio-2-mini · W/ audio historyCER (%): 3.276 -> 3.334No clear effect
add GRPO RL with hybrid reward after SFT (ablation)SFT model · SubjectiveMOS: 3.178 -> 3.576Improved
add GRPO RL with hybrid reward after SFT (ablation)SFT model · W/ audio historyMCLP: -4.725 -> -4.636Improved
add GRPO RL with hybrid reward after SFT (ablation)SFT model · W/o audio historyMCLP: -4.731 -> -4.687Improved
add GRPO RL with hybrid reward after SFT (ablation)SFT model · W/ audio historyCER (%): 3.334 -> 1.130Improved
add GRPO RL with hybrid reward after SFT (ablation)SFT model · W/o audio historyCER (%): 6.306 -> 1.625Improved
add CER content reward to MCLP reward (ablation)Proposed RL · W/ audio historyCER (%): 61.144 -> 1.130Improved
add CER content reward to MCLP reward (ablation)Proposed RL · SubjectiveMOS: 1.145 -> 3.576Improved
add CER content reward to MCLP reward (ablation)Proposed RL · W/ audio historyMCLP: -4.590 -> -4.636Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy