Model merging is a scalable alternative to multi-task training that combines the capabilities of multiple specialised models into a single model. This is particularly attractive for large speech foundation models, which are typically adapted through domain-specific fine-tuning, resulting in multiple customised checkpoints, for which repeating full fine-tuning when new data becomes available is computationally prohibitive. In this work, we study model merging for multi-domain ASR and benchmark 11 merging algorithms for 10 European Portuguese domains, evaluating in-domain accuracy, robustness under distribution shift, as well as English and multilingual performance. We further propose BoostedTSV-M, a new merging algorithm based on TSV-M that mitigates rank collapse via singular-value boosting and improves numerical stability. Overall, our approach outperforms full fine-tuning on European Portuguese while preserving out-of-distribution generalisation in a single model.
@article{arxiv.2603.05354,
title = {Exploring the potential and limitations of Model Merging for Multi-Domain Adaptation in ASR},
author = {Carlos Carvalho and Francisco Teixeira and Thomas Rolland and Alberto Abad},
journal= {arXiv preprint arXiv:2603.05354},
year = {2026}
}
Comments
submitted for review for INTERSPEECH2026 conference