# Large scale weakly and semi-supervised learning for low-resource video ASR

arXiv: https://arxiv.org/abs/2005.07850 · HTML: https://mlsanity.com/paper/2005.07850/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 16 May 2020.

## Built on by (most cited first, 7 of 7)

- [The People's Speech: A Large-Scale Diverse English Speech Recognition Dataset for Commercial Usage](https://mlsanity.com/paper/2111.09344/) (2021)
- [Kaizen: Continuously improving teacher using Exponential Moving Average for semi-supervised speech recognition](https://mlsanity.com/paper/2106.07759/) (2021)
- [Contrastive Semi-supervised Learning for ASR](https://mlsanity.com/paper/2103.05149/) (2021)
- [Low Resource German ASR with Untranscribed Data Spoken by Non-native Children -- INTERSPEECH 2021 Shared Task SPAPL System](https://mlsanity.com/paper/2106.09963/) (2021)
- [Word Order Does Not Matter For Speech Recognition](https://mlsanity.com/paper/2110.05994/) (2021)
- [Semi-supervised transfer learning for language expansion of end-to-end speech recognition models to low-resource languages](https://mlsanity.com/paper/2111.10047/) (2021)
- [Refining Transcripts With TV Subtitles by Prompt-Based Weakly Supervised Training of ASR](https://mlsanity.com/paper/2509.04491/) (2025)

## Abstract (arXiv)

Many semi- and weakly-supervised approaches have been investigated for overcoming the labeling cost of building high quality speech recognition systems. On the challenging task of transcribing social media videos in low-resource conditions, we conduct a large scale systematic comparison between two self-labeling methods on one hand, and weakly-supervised pretraining using contextual metadata on the other. We investigate distillation methods at the frame level and the sequence level for hybrid, encoder-only CTC-based, and encoder-decoder speech recognition systems on Dutch and Romanian languages using 27,000 and 58,000 hours of unlabeled audio respectively. Although all approaches improved upon their respective baseline WERs by more than 8%, sequence-level distillation for encoder-decoder models provided the largest relative WER reduction of 20% compared to the strongest data-augmented supervised baseline.