AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

From Text to Talk: Audio-Language Model Needs Non-Autoregressive Joint Training

Tianqiao Liu, Xueyi Li, Hao Wang, Haoxuan Li, Zhichao Chen, Weiqi Luo, Zitao Liu

ICLR 2026 · Image generation › Autoregressive and latent generation · approach: Neural text-to-speech synthesis · 5 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 batchwise AR & NAR objective mixing (BANOM) (ablation)TtT-3B · Audio-QA LQ.score: 19.87 -> 34.68Improved
add prefix preservation masking (PPM) (ablation)TtT-3B · Audio-QA LQ.score: 22.79 -> 34.68Improved
add stochastic span truncation (SST) (ablation)TtT-3B · Audio-QA LQ.score: 10.20 -> 34.68Improved
add stochastic span truncation (SST) (ablation)TtT-3B · ASR AISHELL-2WER: 25.43 -> 12.53Improved
add stochastic span truncation (SST) (ablation)TtT-3B · ASR AISHELL-1WER: 31.03 -> 13.65Improved
add batchwise AR & NAR objective mixing (BANOM) (ablation)TtT-3B · ASR AISHELL-2WER: 18.58 -> 12.53Improved
add prefix preservation masking (PPM) (ablation)TtT-3B · ASR AISHELL-2WER: 15.63 -> 12.53Improved
add stochastic span truncation (SST) (ablation)TtT-3B · ASR Fen.WER: 62.60 -> 64.31Hurt
replace AR audio generation with hybrid AR text + NAR audio diffusionQwen2.5-3B · Audio-QA LQ.score: 10.00 -> 34.68Improved
replace AR audio generation with hybrid AR text + NAR audio diffusionQwen2.5-3B · ASR AISHELL-2WER: 54.94 -> 12.53Improved
use fully NAR generation for text and audio instead of AR textQwen2.5-3B · ASR AISHELL-2WER: 54.94 -> 212.27Hurt
add large-scale multimodal pretraining before TtT trainingTtT-3B · Audio-QA LQ.score: 34.68 -> 40.07Improved

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