Apple details iterative pseudo-labeling for Mandarin-English code-switching ASR

Apple Machine Learning Research has published a paper detailing a new iterative pseudo-labeling approach for Mandarin-English code-switching Automatic Speech Recognition (ASR). This method leverages unlabeled data to improve ASR performance by generating pseudo-labels, followed by a two-stage bilingual model training process and iterative refinements. The approach has demonstrated significant reductions in Mix Error Rate (MER) on SEAME datasets, achieving 6.35% on devman and 8.29% on devsge. AI

RANK_REASON The cluster contains a research paper published by Apple's Machine Learning Research division on a novel approach to Automatic Speech Recognition. [lever_c_demoted from research: ic=1 ai=1.0]

Read on Apple Machine Learning Research →

• Apple Inc.
• Association for Computational Linguistics
• Cakra Wardhana
• Mandarin-English Code-Switching ASR
• Qu Yang
• Tim Nguyen

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →