AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

TTS Can Speak in Any Style with Any Voice

Dekun Chen, Xueyao Zhang, Yuancheng Wang, Kenan Dai, Li Ma, Zhizheng Wu

ICLR 2026 · Image generation › Autoregressive and latent generation · approach: Neural text-to-speech synthesis

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 S1 multi-modality DPO before S3 (vs S3 only) (ablation)FlexiVoice PPT · InstructTTSEval-ENAvg. accuracy: 72.3 -> 79.3Improved
run S1 then S2 then S3 rather than S3 -> S1 -> S2 (ablation)FlexiVoice PPT · InstructTTSEval-ENAvg. accuracy: 74.8 -> 79.3Improved
train S2 and S3 progressively instead of jointly (ablation)FlexiVoice PPT · Emotion decoupling set (EN)Decoupling Avg.: 84.1 -> 88.7Improved
train S2 and S3 progressively instead of jointly (ablation)FlexiVoice PPT · InstructTTSEval-ENAvg. accuracy: 75.5 -> 79.3Improved
add S1 multi-modality DPO to base model (ablation)FlexiVoice-Base · Emotion decoupling set (EN)Decoupling Avg.: +28.3Improved
add S2 decoupling GRPO after S1 (ablation)FlexiVoice after S1 · Emotion decoupling set (EN)Decoupling Avg.: 83.2 -> 88.5Improved
add S3 instruction GRPO after S1 and S2 (ablation)FlexiVoice after S1+S2 · InstructTTSEval-ENAvg. accuracy: 71.7 -> 79.3 (+7.6)Improved
add S3 instruction GRPO after S1 and S2 (ablation)FlexiVoice after S1+S2 · Emotion decoupling set (EN)Decoupling Avg.: 88.5 -> 88.7No clear effect
use speaker verification reward instead of speaker similarity reward in decoupling GRPO (ablation)FlexiVoice S2 · Emotion decoupling set TR-easy/TR-hard (EN)ACC-IImproved
apply full progressive post-training to pretrained modelFlexiVoice-Base · InstructTTSEval-ENAvg. accuracy: 66.4 -> 79.3Improved
apply full progressive post-training to pretrained modelFlexiVoice-Base · Emotion TR-hard (EN)ACC-I: 48.8 -> 78.2Improved
apply progressive post-training to pretrained modelFlexiVoice-Base · Emotion TO-easy (EN)WER: 5.01 -> 5.99Hurt

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