SSVD: Structured SVD for Parameter-Efficient Fine-Tuning and Benchmarking under Domain Shift in ASR
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866918134911336448 |
|---|---|
| author | Wang, Pu Watanabe, Shinji Van hamme, Hugo |
| author_facet | Wang, Pu Watanabe, Shinji Van hamme, Hugo |
| contents | Parameter-efficient fine-tuning (PEFT) has emerged as a scalable solution for adapting large foundation models. While low-rank adaptation (LoRA) is widely used in speech applications, its state-of-the-art variants, e.g., VeRA, DoRA, PiSSA, and SVFT, are developed mainly for language and vision tasks, with limited validation in speech. This work presents the first comprehensive integration and benchmarking of these PEFT methods within ESPnet. We further introduce structured SVD-guided (SSVD) fine-tuning, which selectively rotates input-associated right singular vectors while keeping output-associated vectors fixed to preserve semantic mappings. This design enables robust domain adaptation with minimal trainable parameters and improved efficiency. We evaluate all methods on domain-shifted speech recognition tasks, including child speech and dialectal variation, across model scales from 0.1B to 2B. All implementations are released in ESPnet to support reproducibility and future work. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_02830 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | SSVD: Structured SVD for Parameter-Efficient Fine-Tuning and Benchmarking under Domain Shift in ASR Wang, Pu Watanabe, Shinji Van hamme, Hugo Computation and Language Audio and Speech Processing Parameter-efficient fine-tuning (PEFT) has emerged as a scalable solution for adapting large foundation models. While low-rank adaptation (LoRA) is widely used in speech applications, its state-of-the-art variants, e.g., VeRA, DoRA, PiSSA, and SVFT, are developed mainly for language and vision tasks, with limited validation in speech. This work presents the first comprehensive integration and benchmarking of these PEFT methods within ESPnet. We further introduce structured SVD-guided (SSVD) fine-tuning, which selectively rotates input-associated right singular vectors while keeping output-associated vectors fixed to preserve semantic mappings. This design enables robust domain adaptation with minimal trainable parameters and improved efficiency. We evaluate all methods on domain-shifted speech recognition tasks, including child speech and dialectal variation, across model scales from 0.1B to 2B. All implementations are released in ESPnet to support reproducibility and future work. |
| title | SSVD: Structured SVD for Parameter-Efficient Fine-Tuning and Benchmarking under Domain Shift in ASR |
| topic | Computation and Language Audio and Speech Processing |
| url | https://arxiv.org/abs/2509.02830 |