cdli
/

whisper-small-Swahili_finetuned_small_CV20

Automatic Speech Recognition

Model card Files Files and versions

trained on: cdli/common_voice_swahili_small

performance (WER):

dev set: 0.261 (pre-tuning: 0.849)
test set: 0.295 (pre-tuning: 0.856)

Downloads last month: 9

Safetensors

Model size

0.2B params

Tensor type

F32

·