metadata

language: en
datasets:
  - librispeech_asr
tags:
  - speech
  - audio
  - automatic-speech-recognition
license: apache-2.0

Distil-wav2vec2

This model is a distilled version of the wav2vec2 model (https://arxiv.org/pdf/2006.11477.pdf). This model is 4 times smaller and 3 times faster than the original wav2vec2 large model.

Evaluation results

When used with a light tri-gram language model head, this model achieves the following results :

Dataset	WER
Librispeech-clean	0.127

Usage

notebook (google colab) at https://github.com/OthmaneJ/distil-wav2vec2