SPaRLoRA: Spectral-Phase Residual Initialization for LoRA in Low-Resource ASR
Open Access
- 16 November 2025
- journal article
- research article
- Published by MDPI AG in Electronics
- Vol. 14 (22) , 4466
Abstract
Parameter-efficient fine-tuning (PEFT) methods like Low-Rank Adaptation (LoRA) are widely used to adapt large pre-trained models under limited resources, yet they often underperform full fine-tuning in low-resource automatic speech recognition (ASR). This gap stems partly from initialization strategies that ignore speech signals’ inherent spectral-phase structure. Unlike SVD/QR-based approaches (PiSSA, OLoRA) that construct mathematically optimal but signal-agnostic subspaces, we propose SPaRLoRA (Spectral-Phase Residual LoRA), which leverages Discrete Fourier Transform (DFT) bases to create speech-aware low-rank adapters. SPaRLoRA explicitly incorporates both magnitude and phase information by concatenating real and imaginary parts of DFT basis vectors, and applies residual correction to focus learning exclusively on components unexplained by the spectral subspace. Evaluated on a 200-h Sichuan dialect ASR benchmark, SPaRLoRA achieves a 2.1% relative character error rate reduction over standard LoRA, outperforming variants including DoRA, PiSSA, and OLoRA. Ablation studies confirm the individual and complementary benefits of spectral basis, phase awareness, and residual correction. Our work demonstrates that signal-structure-aware initialization significantly enhances parameter-efficient fine-tuning for low-resource ASR without architectural changes or added inference cost.Keywords
This publication has 22 references indexed in Scilit:
- Implementation of a Whisper Architecture-Based Turkish Automatic Speech Recognition (ASR) System and Evaluation of the Effect of Fine-Tuning with a Low-Rank Adaptation (LoRA) Adapter on Its PerformanceElectronics, 2024
- Bitune: Bidirectional Instruction-TuningarXiv, 2024
- DoRA: Weight-Decomposed Low-Rank AdaptationarXiv, 2024
- Parameter-efficient fine-tuning of Whisper for low-resource speech recognition2024
- GPT understands, tooAI Open, 2024
- AdaLoRA: Adaptive Budget Allocation for Parameter-Efficient Fine-TuningarXiv, 2023
- A Rank Stabilization Scaling Factor for Fine-Tuning with LoRAarXiv, 2023
- Fine-grained activity classification in assembly based on multi-visual modalitiesJournal of Intelligent Manufacturing, 2023
- Paraformer: Fast and Accurate Parallel Transformer for Non-autoregressive End-to-End Speech RecognitionarXiv, 2023
- Prefix-Tuning: Optimizing Continuous Prompts for GenerationarXiv, 2021