Published on Oct 5
Authors:
,
,
,
,
,
Abstract
Drax, a discrete flow matching framework for ASR, achieves state-of-the-art recognition accuracy with improved efficiency by constructing an audio-conditioned probability path.
AI-generated summary
Diffusion and flow-based non-autoregressive (NAR) models have shown strong promise in large language modeling, however, their potential for automatic speech recognition (ASR) remains largely unexplored. We propose Drax, a discrete flow matching framework for ASR that enables efficient parallel decoding. To better align training with inference, we construct an [audio-conditioned probability pa…
Published on Oct 5
Authors:
,
,
,
,
,
Abstract
Drax, a discrete flow matching framework for ASR, achieves state-of-the-art recognition accuracy with improved efficiency by constructing an audio-conditioned probability path.
AI-generated summary
Diffusion and flow-based non-autoregressive (NAR) models have shown strong promise in large language modeling, however, their potential for automatic speech recognition (ASR) remains largely unexplored. We propose Drax, a discrete flow matching framework for ASR that enables efficient parallel decoding. To better align training with inference, we construct an audio-conditioned probability path that guides the model through trajectories resembling likely intermediate inference errors, rather than direct random noise to target transitions. Our theoretical analysis links the generalization gap to divergences between training and inference occupancies, controlled by cumulative velocity errors, thereby motivating our design choice. Empirical evaluation demonstrates that our approach attains recognition accuracy on par with state-of-the-art speech models while offering improved accuracy-efficiency trade-offs, highlighting discrete flow matching as a promising direction for advancing NAR ASR.
Community
We propose Drax, a non-autoregressive ASR model using discrete flow matching that includes an audio-conditioned intermediate distribution to better match inference dynamics. Drax achieves accuracy comparable to state-of-the-art autoregressive models while offering better control over the accuracy-efficiency trade-off point.
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Models citing this paper 1
aiola/drax-v1 Automatic Speech Recognition • Updated 26 days ago • 32 • 2
Datasets citing this paper 0
No dataset linking this paper
Cite arxiv.org/abs/2510.04162 in a dataset README.md to link it from this page.
Spaces citing this paper 0
No Space linking this paper
Cite arxiv.org/abs/2510.04162 in a Space README.md to link it from this page.