# Distillation and Pruning for Scalable Self-Supervised Representation-Based Speech Quality Assessment

arXiv: https://arxiv.org/abs/2502.05356 · HTML: https://mlsanity.com/paper/2502.05356/

## Key facts

- Cited by 13 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 7 Feb 2025.

## Built on by (most cited first, 13 of 13)

- [Towards Real-Time Generative Speech Restoration with Flow-Matching](https://mlsanity.com/paper/2510.16997/) (2025)
- [Real-Time Streamable Generative Speech Restoration with Flow Matching](https://mlsanity.com/paper/2512.19442/) (2025)
- [Shallow Flow Matching for Coarse-to-Fine Text-to-Speech Synthesis](https://mlsanity.com/paper/2505.12226/) (2025)
- [Diffusion Buffer for Online Generative Speech Enhancement](https://mlsanity.com/paper/2510.18744/) (2025)
- [The Limits of Reference-Free Speech Quality Metrics as Evaluators and Rewards on Modern Text-to-Speech](https://mlsanity.com/paper/2609.13150/) (2026)
- [Real-Time Streaming Mel Vocoding with Generative Flow Matching](https://mlsanity.com/paper/2509.15085/) (2025)
- [Are Modern Speech Enhancement Systems Vulnerable to Adversarial Attacks?](https://mlsanity.com/paper/2509.21087/) (2025)
- [A Fast Solver for Interpolating Stochastic Differential Equation Diffusion Models for Speech Restoration](https://mlsanity.com/paper/2603.09508/) (2026)
- [Real-time Speech Restoration using Data Prediction Mean Flows](https://mlsanity.com/paper/2605.16251/) (2026)
- [CutClean: Neural Network Pruning for Privacy-Preserving Inference](https://mlsanity.com/paper/2608.13773/) (2026)
- [DriftSE: Speech Enhancement with Generative Drifting](https://mlsanity.com/paper/2609.12252/) (2026)
- [Livin' on a Prior: Likelihood Score Approximation for Inverse Problems](https://mlsanity.com/paper/2609.34446/) (2026)
- [Perceptual Quality Loss or Loss of Perceptual Quality?](https://mlsanity.com/paper/2609.35054/) (2026)

## Abstract (arXiv)

In this paper, we investigate distillation and pruning methods to reduce model size for non-intrusive speech quality assessment based on self-supervised representations. Our experiments build on XLS-R-SQA, a speech quality assessment model using wav2vec 2.0 XLS-R embeddings. We retrain this model on a large compilation of mean opinion score datasets, encompassing over 100,000 labeled clips. For distillation, using this model as a teacher, we generate pseudo-labels on unlabeled degraded speech signals and train student models of varying sizes. For pruning, we use a data-driven strategy. While data-driven pruning performs better at larger model sizes, distillation on unlabeled data is more effective for smaller model sizes. Distillation can halve the gap between the baseline's correlation with ground-truth MOS labels and that of the XLS-R-based teacher model, while reducing model size by two orders of magnitude compared to the teacher model.