Reproduce “Acoustic domain shift in spoken language identification from systematic domain generalization evaluation to real-world application” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Francois Derrida, Raphaël Duroselle, Thomas Courtat, Jean-François Bonastre
The authors create a controlled testbed for spoken language identification, comparing training strategies across specified acoustic corruptions. In the CommonVoice four-language task, a wav2vec2-based model trained only on clean speech reached 95.3% accuracy on clean audio but 73.6% on wind-corrupted audio. This measures performance under these conditions, not every real-world recording shift.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.