README.md

metadata

license: mit
datasets:
  - librispeech_asr
language:
  - en
metrics:
  - wer
library_name: transformers

Wav2Vec2 XLS-R-300m Finetuning using 1hr of Librispeech

XLS-R 300m pretrained model found here is finetuned using 1hr of LibriSpeech Train-Clean-100 subset sampled at random.

XLS-R paper

Training Code is available on this colab notebook. (The notebook is inspired by this notebook)