fleurs-vi vn.train.48c33a4427d6cec339b5055b58cd938205b69510c6fba1602421b10a581a09f7
fleurs-vi vn.train.48c33a4427d6cec339b5055b58cd938205b69510c6fba1602421b10a581a09f7: train task in FineEnvs/fleurs-asr-env, an RL environment on the Hugging Face Hub. Transcribe this Vietnamese speech exactly as spoken, in Vietnamese, preserving capitalization, punctuation, and numerals. Return…
The task
Transcribe this Vietnamese speech exactly as spoken, in Vietnamese, preserving capitalization, punctuation, and numerals. Return only the transcription.
Task details
- language
- vi_vn
- language name
- Vietnamese
- family
- verbatim_transcription
- mime
- audio/wav
- duration seconds
- 14.58
- sampling rate
- 16000
- media ready
- False