---
license: afl-3.0
---

FineTuned wav2vec2 large 960H lv60 self pre-trained facebook model on 72 Hours of MI Diaries Data

WER 13 % -> 9.7% on 20 min test set of MI Diaries audio clips (https://mi-diaries.org/)


### Usage ###
model = Wav2Vec2ForCTC.from_pretrained("caurdy/wav2vec2-large-960h-lv60-self_MIDIARIES_72H_FT")

processor = Wav2Vec2Processor.from_pretrained("caurdy/wav2vec2-large-960h-lv60-self_MIDIARIES_72H_FT")