Fithubert: Going Thinner And Deeper For Knowledge Distillation Of Speech Self-supervised Learning

Abstract

Large-scale speech self-supervised learning (SSL) has emerged to the main field of speech processing, however, the problem of computational cost arising from its vast size makes a high entry barrier to academia. In addition, existing distillation techniques of speech SSL models compress the model by reducing layers, which induces performance degradation in linguistic pattern recognition tasks such as phoneme recognition (PR). In this paper, we propose FitHuBERT, which makes thinner in dimension throughout almost all model components and deeper in layer compared to prior speech SSL distillation works. Moreover, we employ a time-reduction layer to speed up inference time and propose a method of hint-based distillation for less performance degradation. Our method reduces the model to 23.8% in size and 35.9% in inference time compared to HuBERT. Also, we achieve 12.1% word error rate and 13.3% phoneme error rate on the SUPERB benchmark which is superior than prior work.

Stats

• citations28
• S2 citations—
• github stars0
• HF likes0
• heat score10.97
• arxiv keylee2022fithubert

Related papers

• Distilhubert: Speech Representation Learning By Layer-wise Distillation Of Hidden-unit BERT (2021)15.06
• Fast-hubert: An Efficient Training Framework For Self-supervised Speech Representation Learning (2023)0.00
• SKILL: Similarity-aware Knowledge Distillation For Speech Self-supervised Learning (2024)3.58
• Recycle-and-distill: Universal Compression Strategy For Transformer-based Speech SSL Models With Attention Map Reusing And Masking Distillation (2023)5.84
• Star: Distilling Speech Temporal Relation For Lightweight Speech Self-supervised Learning Models (2023)5.24
• Deep Versus Wide: An Analysis Of Student Architectures For Task-agnostic Knowledge Distillation Of Self-supervised Speech Models (2022)9.23
• Pushing The Limits Of Unsupervised Unit Discovery For SSL Speech Representation (2023)6.34
• One-step Knowledge Distillation And Fine-tuning In Using Large Pre-trained Self-supervised Learning Models For Speaker Verification (2023)7.81