metadata
language:
- en
tags:
- automatic-speech-recognition
datasets:
- LIUM/tedlium
license: cc-by-4.0
metrics:
- name: Dev WER
type: wer
value: 9
- name: Test WER
type: wer
value: 6.4
wav2vec2-2-bart-large-tedlium
This model is a sequence-2-sequence (seq2seq) model trained on the TEDLIUM corpus (release 3).
It combines a speech encoder with a text decoder to perform automatic speech recognition. The encoder weights are initialised with the Wav2Vec2 LV-60k checkpoint from @facebook. The decoder weights are initialised with the Bart large checkpoint from @facebook.