Alexei Baevski

Latest

vq-wav2vec: Self-Supervised Learning of Discrete Speech Representations

Learning discrete representations of speech yields state-of-the-art recognition performance on TIMIT and WSJ.
Alexei Baevski, Steffen Schneider, Michael Auli
PDF Code Video

wav2vec: Unsupervised Pre-training for Speech Recognition

Contrastive pre-training on speech at scale reduces the need for labeled data.
Steffen Schneider, Alexei Baevski, Ronan Collobert, Michael Auli
PDF Code