# DistillW2V2: A Small and Streaming Wav2vec 2.0 Based ASR Model

arXiv: https://arxiv.org/abs/2303.09278 · HTML: https://mlsanity.com/paper/2303.09278/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- Placed in Speech and audio distillation, one of 12 directions in Knowledge distillation.
- Builds on 17 of its 25 references that are in the map.
- First posted on arXiv 16 Mar 2023.

## Placed in

- Knowledge distillation / [Speech and audio distillation](https://mlsanity.com/distill/speech-and-audio-distillation/)

## Builds on (most cited first, 17 of 17)

- [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://mlsanity.com/paper/2006.11477/) (2020)
- [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://mlsanity.com/paper/2110.13900/) (2021)
- [TinyBERT: Distilling BERT for Natural Language Understanding](https://mlsanity.com/paper/1909.10351/) (2019)
- [GigaSpeech: An Evolving, Multi-domain ASR Corpus with 10,000 Hours of Transcribed Audio](https://mlsanity.com/paper/2106.06909/) (2021)
- [W2v-BERT: Combining Contrastive Learning and Masked Language Modeling for Self-Supervised Speech Pre-Training](https://mlsanity.com/paper/2108.06209/) (2021)
- [Unsupervised Speech Recognition](https://mlsanity.com/paper/2105.11084/) (2021)
- [DistilHuBERT: Speech Representation Learning by Layer-wise Distillation of Hidden-unit BERT](https://mlsanity.com/paper/2110.01900/) (2021)
- [Developing Real-time Streaming Transformer Transducer for Speech Recognition on Large-scale Dataset](https://mlsanity.com/paper/2010.11395/) (2020)
- [LightHuBERT: Lightweight and Configurable Speech Representation Learning with Once-for-All Hidden-Unit BERT](https://mlsanity.com/paper/2203.15610/) (2022)
- [Efficient Knowledge Distillation for RNN-Transducer Models](https://mlsanity.com/paper/2011.06110/) (2020)
- [FitHuBERT: Going Thinner and Deeper for Knowledge Distillation of Speech Self-Supervised Learning](https://mlsanity.com/paper/2207.00555/) (2022)
- [Shrinking Bigfoot: Reducing wav2vec 2.0 footprint](https://mlsanity.com/paper/2103.15760/) (2021)
- [Knowledge Distillation for Neural Transducers from Large Self-Supervised Pre-trained Models](https://mlsanity.com/paper/2110.03334/) (2021)
- [Improving Hybrid CTC/Attention End-to-end Speech Recognition with Pretrained Acoustic and Language Model](https://mlsanity.com/paper/2112.07254/) (2021)
- [Improving Streaming Transformer Based ASR Under a Framework of Self-supervised Learning](https://mlsanity.com/paper/2109.07327/) (2021)
- [Comparing CTC and LFMMI for out-of-domain adaptation of wav2vec 2.0 acoustic model](https://mlsanity.com/paper/2104.02558/) (2021)
- [Two-Pass End-to-End ASR Model Compression](https://mlsanity.com/paper/2201.02741/) (2022)

## Built on by (most cited first, 7 of 7)

- [Simul-Whisper: Attention-Guided Streaming Whisper with Truncation Detection](https://mlsanity.com/paper/2406.10052/) (2024)
- [Effective and Efficient Mixed Precision Quantization of Speech Foundation Models](https://mlsanity.com/paper/2501.03643/) (2025)
- [One-pass Multiple Conformer and Foundation Speech Systems Compression and Quantization Using An All-in-one Neural Model](https://mlsanity.com/paper/2406.10160/) (2024)
- [AdaKD: Dynamic Knowledge Distillation of ASR models using Adaptive Loss Weighting](https://mlsanity.com/paper/2405.08019/) (2024)
- [Effective and Efficient One-pass Compression of Speech Foundation Models Using Sparsity-aware Self-pinching Gates](https://mlsanity.com/paper/2505.22608/) (2025)
- [Dynamic Fusion Multimodal Network for SpeechWellness Detection](https://mlsanity.com/paper/2508.18057/) (2025)
- [Towards Data-free and Training-free Compression for Speech Foundation Models Using Parameter Clustering](https://mlsanity.com/paper/2606.11836/) (2026)

## Abstract (arXiv)

Wav2vec 2.0 (W2V2) has shown impressive performance in automatic speech recognition (ASR). However, the large model size and the non-streaming architecture make it hard to be used under low-resource or streaming scenarios. In this work, we propose a two-stage knowledge distillation method to solve these two problems: the first step is to make the big and non-streaming teacher model smaller, and the second step is to make it streaming. Specially, we adopt the MSE loss for the distillation of hidden layers and the modified LF-MMI loss for the distillation of the prediction layer. Experiments are conducted on Gigaspeech, Librispeech, and an in-house dataset. The results show that the distilled student model (DistillW2V2) we finally get is 8x faster and 12x smaller than the original teacher model. For the 480ms latency setup, the DistillW2V2's relative word error rate (WER) degradation varies from 9% to 23.4% on test sets, which reveals a promising way to extend the W2V2's application scope.