# Application of Knowledge Distillation to Multi-task Speech Representation Learning

arXiv: https://arxiv.org/abs/2210.16611 · HTML: https://mlsanity.com/paper/2210.16611/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Speech and audio distillation, one of 12 directions in Knowledge distillation.
- Builds on 22 of its 29 references that are in the map.
- First posted on arXiv 29 Oct 2022.

## Placed in

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

## Builds on (most cited first, 22 of 22)

- [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://mlsanity.com/paper/2006.11477/) (2020)
- [HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units](https://mlsanity.com/paper/2106.07447/) (2021)
- [Knowledge Distillation: A Survey](https://mlsanity.com/paper/2006.05525/) (2020)
- [Speech Commands: A Dataset for Limited-Vocabulary Speech Recognition](https://mlsanity.com/paper/1804.03209/) (2018)
- [VoxCeleb: a large-scale speaker identification dataset](https://mlsanity.com/paper/1706.08612/) (2017)
- [SUPERB: Speech processing Universal PERformance Benchmark](https://mlsanity.com/paper/2105.01051/) (2021)
- [vq-wav2vec: Self-Supervised Learning of Discrete Speech Representations](https://mlsanity.com/paper/1910.05453/) (2019)
- [Layer-wise Analysis of a Self-supervised Speech Representation Model](https://mlsanity.com/paper/2107.04734/) (2021)
- [Mockingjay: Unsupervised Speech Representation Learning with Deep Bidirectional Transformer Encoders](https://mlsanity.com/paper/1910.12638/) (2019)
- [TERA: Self-Supervised Learning of Transformer Encoder Representation for Speech](https://mlsanity.com/paper/2007.06028/) (2020)
- [An Unsupervised Autoregressive Model for Speech Representation Learning](https://mlsanity.com/paper/1904.03240/) (2019)
- [Emotion Recognition from Speech Using Wav2vec 2.0 Embeddings](https://mlsanity.com/paper/2104.03502/) (2021)
- [DistilHuBERT: Speech Representation Learning by Layer-wise Distillation of Hidden-unit BERT](https://mlsanity.com/paper/2110.01900/) (2021)
- [A Fine-tuned Wav2vec 2.0/HuBERT Benchmark For Speech Emotion Recognition, Speaker Verification and Spoken Language Understanding](https://mlsanity.com/paper/2111.02735/) (2021)
- [Large-scale Self-Supervised Speech Representation Learning for Automatic Speaker Verification](https://mlsanity.com/paper/2110.05777/) (2021)
- [Speech Emotion Recognition using Self-Supervised Features](https://mlsanity.com/paper/2202.03896/) (2022)
- [LightHuBERT: Lightweight and Configurable Speech Representation Learning with Once-for-All Hidden-Unit BERT](https://mlsanity.com/paper/2203.15610/) (2022)
- [Fine-tuning wav2vec2 for speaker recognition](https://mlsanity.com/paper/2109.15053/) (2021)
- [An Exploration of Self-Supervised Pretrained Representations for End-to-End Speech Recognition](https://mlsanity.com/paper/2110.04590/) (2021)
- [Deep Representation Learning in Speech Processing: Challenges, Recent Advances, and Future Trends](https://mlsanity.com/paper/2001.00378/) (2020)
- [Shrinking Bigfoot: Reducing wav2vec 2.0 footprint](https://mlsanity.com/paper/2103.15760/) (2021)
- [Multi-task Voice Activated Framework using Self-supervised Learning](https://mlsanity.com/paper/2110.01077/) (2021)

## Abstract (arXiv)

Model architectures such as wav2vec 2.0 and HuBERT have been proposed to learn speech representations from audio waveforms in a self-supervised manner. When they are combined with downstream tasks such as keyword spotting and speaker verification, they provide state-of-the-art performance. However, these models use a large number of parameters, the smallest version of which has 95 million parameters. This constitutes a challenge for edge AI device deployments. In this paper, we investigate the application of knowledge distillation to speech representation learning (SRL) models followed by joint fine-tuning with multiple downstream voice-activated tasks. In our experiments on two such tasks, our approach results in nearly 75% reduction in model size while suffering only 0.1% accuracy and 0.9% equal error rate degradation compared to the full-size model. In addition, we show that fine-tuning the SRL models results in a significant performance boost compared to using frozen SRL models.