metadata
language:
- ru
license: apache-2.0
tags:
- automatic-speech-recognition
- mozilla-foundation/common_voice_8_0
- ru
- robust-speech-event
datasets:
- mozilla-foundation/common_voice_8_0
model-index:
- name: XLS-R-1B - Russian
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Common Voice 8
type: mozilla-foundation/common_voice_8_0
args: ru
metrics:
- name: Test WER
type: wer
value: 9.87
- name: Test CER
type: cer
value: 2.32
- name: Test WER (+LM)
type: wer
value: null
- name: Test CER (+LM)
type: cer
value: null
XLS-R-1B-RUSSIAN
This model is a fine-tuned version of facebook/wav2vec2-xls-r-1b on the MOZILLA-FOUNDATION/COMMON_VOICE_8_0 - RU dataset.
Evaluation Commands
- To evaluate on
mozilla-foundation/common_voice_8_0
with splittest
python eval.py --model_id jonatasgrosman/wav2vec2-xls-r-1b-russian --dataset mozilla-foundation/common_voice_8_0 --config ru --split test
- To evaluate on
speech-recognition-community-v2/dev_data
python eval.py --model_id jonatasgrosman/wav2vec2-xls-r-1b-russian --dataset speech-recognition-community-v2/dev_data --config ru --split validation --chunk_length_s 5.0 --stride_length_s 1.0