AI Paper Atlas

Research map / Label-efficient and robust learning / Self-supervised and contrastive

Pay Attention to CTC: Fast and Robust Pseudo-Labelling for Unified Speech Recognition

Alexandros Haliassos, Rodrigo Mira, Stavros Petridis

ICLR 2026 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Self-supervised speech and brain decoding · 1 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace USR autoregressive pseudo-labelling with CTC-driven teacher forcing plus mixed sampling (USR 2.0)USR (Large, LRS3+Vox2) · LRS3 V (low-resource)WER (%): 26.9 -> 23.7Improved
USR 2.0 vs USR at Base with LRS3 onlyUSR (Base) · LRS3 V (low-resource)WER (%): 36.0 -> 36.2No clear effect
USR 2.0 vs USR under babble noiseUSR (Base) · LRS3 ASR noisy, avgWER (%): 43.3 -> 39.3Improved
add CTC PL supervision to the decoder in CTC-driven mode (ablation)USR 2.0 · AVSpeech OOD AVSRWER (%): 35.1 -> 24.2Improved
add attention-PL supervision to the decoder in CTC-driven mode (ablation)USR 2.0 · LRS3 AVSR (ID)WER (%): 3.6 -> 3.2Improved
supervise CTC head with attention PLs in CTC-driven mode (ablation)USR 2.0 · LRS3 / AVSpeech AVSRWER (%): 3.2/24.2 -> 3.2/24.3No clear effect
add attention PLs to CTC head in AR mode (ablation)USR 2.0 · AVSpeech OOD AVSRWER (%): 45.1 -> 40.1Improved
add CTC PLs to the CTC head in AR mode alongside attention PLs (ablation)USR 2.0 · AVSpeech OOD AVSRWER (%): 52.3 -> 40.1Improved
increase AR mode sampling probability toward 1 (ablation)USR 2.0 · LRS3 ID vs AVSpeech OOD AVSRWER (%)Mixed
adaptive (linear or cosine) AR sampling schedule instead of constant 0.5 (ablation)USR 2.0 · AVSpeech OOD AVSRWER (%): 25.0 -> 24.8 (linear), 25.4 (cosine)No clear effect
apply CTC merge and collapse to USR pseudo-labels alone (ablation)USR · AVSpeech OOD AVSRWER (%): 34.7 -> 36.9Hurt
CTC-driven teacher forcing instead of AR pseudo-labellingUSR · VSR training timewall-clock training time: ~2x fasterImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy