AI Paper Atlas

Research map / Large language models / Alignment and preferences

SpeechJudge: Towards Human-Level Judgment for Speech Naturalness

Xueyao Zhang, Chaoren Wang, Huan Liao, Ziniu Li, Yuancheng Wang, Li Wang, Dongya Jia, Yuanzhe Chen, Xiulin LI, Zhuo Chen, Zhizheng Wu

ICLR 2026 · Large language models › Alignment and preferences · approach: LLM-as-judge and feedback data · 24 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
Use generative reward model (SFT+RL) instead of Bradley-Terry reward modelQwen2.5-Omni-7B judge · SpeechJudge-Evalaccuracy: 72.7 -> 77.2Improved
Add GRPO RL stage after SFT (ablation)SpeechJudge-GRM · SpeechJudge-Evalaccuracy: 75.3 -> 77.2Improved
Add majority voting over 10 outputs (ablation)SpeechJudge-GRM (SFT) · SpeechJudge-Evalaccuracy: 75.3 -> 77.6Improved
Add majority voting over 10 outputs (ablation)SpeechJudge-GRM (SFT+RL) · SpeechJudge-Evalaccuracy: 77.2 -> 79.4Improved
Use CoT prompt instead of plain prompt (ablation)Gemini-2.5-Flash · SpeechJudge-Evalaccuracy: 69.1 -> 70.5Improved
Use CoT prompt instead of plain prompt (ablation)Qwen2.5-Omni-7B · SpeechJudge-Evalaccuracy: 60.6 -> 52.9Hurt
Use CoT prompt instead of plain prompt (ablation)Kimi-Audio-7B-Instruct · SpeechJudge-Evalaccuracy: 67.0 -> 66.5No clear effect
Add 2-shot exemplars to the prompt (ablation)Qwen2.5-Omni-7B · SpeechJudge-Evalaccuracy: 60.6 -> 48.2Hurt
Add GRPO RL stage after SFT (ablation)SpeechJudge-GRM · Human vs SeedTTS clone (OOD)accuracy: 40.8 -> 62.4Improved
Use SpeechJudge-GRM (SFT+RL) instead of BTRMQwen2.5-Omni-7B judge · Human vs SeedTTS clone (OOD)accuracy: 50.4 -> 62.4Improved
Post-train Qwen2.5-0.5B-TTS with SpeechJudge-GRM reward instead of INTP DPO dataQwen2.5-0.5B-TTS · SeedTTS-Eval and Amphion-TTS-Evalnaturalness CMOSImproved

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