ML Anthology

Vq-Wav2vec: Self-Supervised Learning of Discrete Speech Representations

Alexei Baevski, Steffen Schneider, Michael Auli
ICLR 2020
/iclr/2020/baevski2020iclr-vqwav2vec/

Abstract

We propose vq-wav2vec to learn discrete representations of audio segments through a wav2vec-style self-supervised context prediction task. The algorithm uses either a gumbel softmax or online k-means clustering to quantize the dense representations. Discretization enables the direct application of algorithms from the NLP community which require discrete inputs. Experiments show that BERT pre-training achieves a new state of the art on TIMIT phoneme classification and WSJ speech recognition.

PDF ICLR Semantic Scholar

Cite

Text

Baevski et al. "Vq-Wav2vec: Self-Supervised Learning of Discrete Speech Representations." International Conference on Learning Representations, 2020.

Markdown

[Baevski et al. "Vq-Wav2vec: Self-Supervised Learning of Discrete Speech Representations." International Conference on Learning Representations, 2020.](https://mlanthology.org/iclr/2020/baevski2020iclr-vqwav2vec/)

BibTeX

@inproceedings{baevski2020iclr-vqwav2vec,
  title     = {{Vq-Wav2vec: Self-Supervised Learning of Discrete Speech Representations}},
  author    = {Baevski, Alexei and Schneider, Steffen and Auli, Michael},
  booktitle = {International Conference on Learning Representations},
  year      = {2020},
  url       = {https://mlanthology.org/iclr/2020/baevski2020iclr-vqwav2vec/}
}