facebook/wav2vec2-large-960h-lv60-self
Automatic Speech Recognition • Updated • 59.6k • 162
Wav2Vec 2.0
A collection for the first release of Wav2Vec 2.0, a speech encoder that learns powerful representations from unlabelled audio data.
Note The Wav2Vec 2.0 "large" model pre-trained on 53k hours of un-labelled audio data from the LibriSpeech and LibriVox (LV) corpora, and fine-tuned on 960 hours of LibriSpeech ASR data. This is the most performant Wav2Vec 2.0 checkpoint from the initial release, obtaining 1.9/3.9% WER on the LibriSpeech test clean/other subsets respectively.
facebook/wav2vec2-large-960h
Automatic Speech Recognition • Updated • 27.3k • 35Note The Wav2Vec 2.0 "large" model pre-trained and fine-tuned on 960 hours of LibriSpeech ASR data.
facebook/wav2vec2-base-960h
Automatic Speech Recognition • 94.4M • Updated • 1.74M • 402Note The Wav2Vec 2.0 "base" model pre-trained and fine-tuned on 960 hours of LibriSpeech ASR data.
facebook/wav2vec2-base-100h
Automatic Speech Recognition • Updated • 391 • 7Note The Wav2Vec 2.0 "base" model pre-trained on 960 hours of un-labelled LibriSpeech ASR data, and fine-tuned on 100 hours of labelled LibriSpeech ASR data.
facebook/wav2vec2-large-lv60
Updated • 5.17k • 14Note The Wav2Vec 2.0 "large" model pre-trained on 53k hours of un-labelled data from the LibriSpeech and LibriVox (LV) corpora.
facebook/wav2vec2-large
Updated • 34k • 10Note The Wav2Vec 2.0 "large" model pre-trained on 960 hours of un-labelled LibriSpeech ASR data.
facebook/wav2vec2-base
Updated • 3.32M • 124Note The Wav2Vec 2.0 "base" model pre-trained on 960 hours of un-labelled LibriSpeech ASR data.
wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations
Paper • 2006.11477 • Published • 10Note The wav2vec 2.0 paper, accepted to NeurIPS 2020.