Progressive Refinement: An Iterative Pseudo-Labeling Approach for Mandarin-English Code-Switching ASR
Apple researchers have developed an iterative pseudo-labeling method to improve automatic speech recognition for Mandarin-English code-switching. By using this approach to generate training data for mixed-language speech, the team aims to overcome performance gaps caused by a lack of diverse, multilingual speech samples.
Covered by 1 source
- AApple Machine Learning Blog↗2d ago