# Speech Meets ELF: Audio Conditional Continuous-Target Diffusion for Speech Recognition and Translation

arXiv: https://arxiv.org/abs/2606.10368 · HTML: https://mlsanity.com/paper/2606.10368/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 8 of its 21 references that are in the map.
- First posted on arXiv 9 Jun 2026.

## Builds on (most cited first, 8 of 8)

- [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://mlsanity.com/paper/1910.10683/) (2019)
- [Robust Speech Recognition via Large-Scale Weak Supervision](https://mlsanity.com/paper/2212.04356/) (2022)
- [Discrete Diffusion Modeling by Estimating the Ratios of the Data Distribution](https://mlsanity.com/paper/2310.16834/) (2023)
- [Prompting Large Language Models with Speech Recognition Abilities](https://mlsanity.com/paper/2307.11795/) (2023)
- [Connecting Speech Encoder and Large Language Model for ASR](https://mlsanity.com/paper/2309.13963/) (2023)
- [The Curse of Multi-Modalities: Evaluating Hallucinations of Large Multimodal Models across Language, Visual, and Audio](https://mlsanity.com/paper/2410.12787/) (2024)
- [Comparing Discrete and Continuous Space LLMs for Speech Recognition](https://mlsanity.com/paper/2409.00800/) (2024)
- [Speech Discrete Tokens or Continuous Features? A Comparative Analysis for Spoken Language Understanding in SpeechLLMs](https://mlsanity.com/paper/2508.17863/) (2025)

## Abstract (arXiv)

Speech-to-text (S2T) systems for recognition (ASR) and translation (S2TT) typically generate discrete text tokens. In contrast, continuous-target language modelling performs generation in a continuous space, yet its potential for S2T remains unexplored. To bridge this gap, we propose ELF-S2T, an audio-conditioned continuous-target generative model for S2T. Built upon the pre-trained Embedded Language Flows (ELF) backbone, ELF-S2T processes speech via a frozen Whisper encoder and a single linear projector, prepending the resulting audio condition to the noisy text latent for in-context, flow-matching denoising. To prevent the model from over-relying on its pre-trained text context, we introduce audio forcing during training, and further amplify the audio condition via classifier-free guidance at inference. Experiments on LibriSpeech and CoVoST2 show that ELF-S2T achieves competitive ASR and S2TT performance. Crucially, our error analysis reveals that, although ASR and S2TT errors look very different on the surface, both stem from the same underlying cause, a close distance confusion in the continuous latent space. This finding naturally aligns with the continuous representation generation paradigm, indicating a common semantic mapping process beneath recognition and translation. Our code and pretrained models are publicly available at https://github.com/Sslnon/ELF-S2T.