CarelessWhisper: Turning Whisper into a Causal Streaming Model

• URL: http://arxiv.org/abs/2508.12301v1
• Date: Sun, 17 Aug 2025 09:32:40 GMT
• Title: CarelessWhisper: Turning Whisper into a Causal Streaming Model
• Authors: Tomer Krichli, Bhiksha Raj, Joseph Keshet,
• Abstract summary: We present an analysis explaining why it is not straightforward to convert an encoder-decoder transformer to a low-latency streaming model.<n>Our proposed method modifies the existing (non-causal) encoder to a causal encoder by fine-tuning both the encoder and decoder.<n> Experiments on low-latency chunk sizes (less than 300 msec) show that our fine-tuned model outperforms existing non-fine-tuned streaming approaches.
• Score: 31.38962687054824
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: Automatic Speech Recognition (ASR) has seen remarkable progress, with models like OpenAI Whisper and NVIDIA Canary achieving state-of-the-art (SOTA) performance in offline transcription. However, these models are not designed for streaming (online or real-time) transcription, due to limitations in their architecture and training methodology. We propose a method to turn the transformer encoder-decoder model into a low-latency streaming model that is careless about future context. We present an analysis explaining why it is not straightforward to convert an encoder-decoder transformer to a low-latency streaming model. Our proposed method modifies the existing (non-causal) encoder to a causal encoder by fine-tuning both the encoder and decoder using Low-Rank Adaptation (LoRA) and a weakly aligned dataset. We then propose an updated inference mechanism that utilizes the fine-tune causal encoder and decoder to yield greedy and beam-search decoding, and is shown to be locally optimal. Experiments on low-latency chunk sizes (less than 300 msec) show that our fine-tuned model outperforms existing non-fine-tuned streaming approaches in most cases, while using a lower complexity. Additionally, we observe that our training process yields better alignment, enabling a simple method for extracting word-level timestamps. We release our training and inference code, along with the fine-tuned models, to support further research and development in streaming ASR.

Related papers

• Voxtral Realtime [134.66962524291424]
  Voxtral Realtime is a streaming automatic speech recognition model.<n>It matches offline transcription quality at sub-second latency.<n>We release the model weights under the Apache 2.0 license.
  arXiv  Detail & Related papers  (2026-02-11T19:17:10Z)
• Stateful Conformer with Cache-based Inference for Streaming Automatic Speech Recognition [20.052245837954175]
  We propose an efficient and accurate streaming speech recognition model based on the FastConformer architecture. We introduce an activation caching mechanism to enable the non-autoregressive encoder to operate autoregressively during inference. A hybrid CTC/RNNT architecture which utilizes a shared encoder with both a CTC and RNNT decoder to boost the accuracy and save computation.
  arXiv  Detail & Related papers  (2023-12-27T21:04:26Z)
• Faster Diffusion: Rethinking the Role of the Encoder for Diffusion Model Inference [95.42299246592756]
  We study the UNet encoder and empirically analyze the encoder features. We find that encoder features change minimally, whereas the decoder features exhibit substantial variations across different time-steps. We validate our approach on other tasks: text-to-video, personalized generation and reference-guided generation.
  arXiv  Detail & Related papers  (2023-12-15T08:46:43Z)
• Semi-Autoregressive Streaming ASR With Label Context [70.76222767090638]
  We propose a streaming "semi-autoregressive" ASR model that incorporates the labels emitted in previous blocks as additional context. Experiments show that our method outperforms the existing streaming NAR model by 19% relative on Tedlium2, 16%/8% on Librispeech-100 clean/other test sets, and 19%/8% on the Switchboard(SWB)/Callhome(CH) test sets.
  arXiv  Detail & Related papers  (2023-09-19T20:55:58Z)
• ED2LM: Encoder-Decoder to Language Model for Faster Document Re-ranking Inference [70.36083572306839]
  This paper proposes a new training and inference paradigm for re-ranking. We finetune a pretrained encoder-decoder model using in the form of document to query generation. We show that this encoder-decoder architecture can be decomposed into a decoder-only language model during inference.
  arXiv  Detail & Related papers  (2022-04-25T06:26:29Z)
• Streaming Align-Refine for Non-autoregressive Deliberation [42.748839817396046]
  We propose a streaming non-autoregressive (non-AR) decoding algorithm to deliberate the hypothesis alignment of a streaming RNN-T model. Our algorithm facilitates a simple greedy decoding procedure, and at the same time is capable of producing the decoding result at each frame with limited right context. Experiments on voice search datasets and Librispeech show that with reasonable right context, our streaming model performs as well as the offline counterpart.
  arXiv  Detail & Related papers  (2022-04-15T17:24:39Z)
• Deliberation of Streaming RNN-Transducer by Non-autoregressive Decoding [21.978994865937786]
  The method performs a few refinement steps, where each step shares a transformer decoder that attends to both text features and audio features. We show that, conditioned on hypothesis alignments of a streaming RNN-T model, our method obtains significantly more accurate recognition results than the first-pass RNN-T.
  arXiv  Detail & Related papers  (2021-12-01T01:34:28Z)
• Unified Streaming and Non-streaming Two-pass End-to-end Model for Speech Recognition [19.971343876930767]
  We present a novel two-pass approach to unify streaming and non-streaming end-to-end (E2E) speech recognition in a single model. Our model adopts the hybrid CTC/attention architecture, in which the conformer layers in the encoder are modified. Experiments on the open 170-hour AISHELL-1 dataset show that, the proposed method can unify the streaming and non-streaming model simply and efficiently.
  arXiv  Detail & Related papers  (2020-12-10T06:54:54Z)
• Cascaded encoders for unifying streaming and non-streaming ASR [68.62941009369125]
  This work presents cascaded encoders for building a single E2E ASR model that can operate in both these modes simultaneously. A single decoder then learns to decode either using the output of the streaming or the non-streaming encoder. Results show that this model achieves similar word error rates (WER) as a standalone streaming model when operating in streaming mode, and obtains 10% -- 27% relative improvement when operating in non-streaming mode.
  arXiv  Detail & Related papers  (2020-10-27T20:59:50Z)
• Conv-Transformer Transducer: Low Latency, Low Frame Rate, Streamable End-to-End Speech Recognition [8.046120977786702]
  Transformer has achieved competitive performance against state-of-the-art end-to-end models in automatic speech recognition (ASR) The original Transformer, with encoder-decoder architecture, is only suitable for offline ASR. We show that this architecture, named Conv-Transformer Transducer, achieves competitive performance on LibriSpeech dataset (3.6% WER on test-clean) without external language models.
  arXiv  Detail & Related papers  (2020-08-13T08:20:02Z)
• Streaming automatic speech recognition with the transformer model [59.58318952000571]
  We propose a transformer based end-to-end ASR system for streaming ASR. We apply time-restricted self-attention for the encoder and triggered attention for the encoder-decoder attention mechanism. Our proposed streaming transformer architecture achieves 2.8% and 7.2% WER for the "clean" and "other" test data of LibriSpeech.
  arXiv  Detail & Related papers  (2020-01-08T18:58:02Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.