Datasets:
				
			
			
	
			
	
		
			
	
		Modalities:
	
	
	
		
	
	Text
	
	
	Formats:
	
	
	
		
	
	parquet
	
	
	Languages:
	
	
	
		
	
	Welsh
	
	
	Size:
	
	
	
	
	10K - 100K
	
	
	Tags:
	
	
	
	
	speech recognition
	
	
	License:
	
	
	
	
	
	
	
metadata
			language:
  - cy
license: cc0-1.0
tags:
  - speech recognition
metrics:
  - wer
  - cer
| test | wer | cer | 
|---|---|---|
| techiaith/banc-trawsgrifiadau-bangor/test | 47.2219 | 24.0401 | 
| cymen-arfor/lleisiau-arfor/test | 56.9979 | 30.4421 | 
| wanasash/corpus-siarad-test-set/segments | 89.882 | 63.119 | 
| techiaith/commonvoice-18-0-cy/test | 50.9921 | 17.9207 | 
| techiaith/commonvoice-18-0-cy-en/test | 81.4001 | 46.1843 | 
| techiaith/YouTube-Subtitles/train | 87.3888 | 55.4715 | 
| techiaith/gweferydd-bbc-ec/pronunciation-corrections | 50.6617 | 22.4682 | 
| techiaith/gweferydd-bbc-ec/pronunciation-additions | 48.0829 | 21.2234 | 

