infinitejoy / automatic-speech-recognition updated 4 years ago

wav2vec2-large-xls-r-300m-welsh

Name: wav2vec2-large-xls-r-300m-welsh
Author: infinitejoy

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the MOZILLA-FOUNDATION/COMMONVOICE70 - CY dataset. It achieves the following results on the evaluation set: - Loss: 0.2650 - Wer: 0.2702

Params

300 M

Context

—

Downloads 30d

567 K

Likes

Commercial use: allowed apache-2.0 Not gated 1 languages View on Hugging Face ↗

Download history

daily snapshots · 36 days

▲ 193 K in the last 30 days (51.6%)

656 K374 K

Aug 17Aug 29Sep 10Sep 21

Specifications

Architecture: Wav2Vec2ForCTC
Parameters: 300 M
Vocabulary: 49
Layers / heads: 24 / 16
Licence: apache-2.0
First seen on the Hub: 2022-03-02
Training datasets: mozilla-foundation/common_voice_7_0
Common Voice 7 (reported): 7.775
Added to our catalog: 2026-08-17

Compare with any automatic-speech-recognition model

Popular comparisons

wav2vec2-large-xlsr-53-japanese 17.7 M whisperkit-coreml 11.1 M speaker-diarization-3.1 8.0 M wav2vec2-large-xlsr-53-portuguese 5.9 M speaker-diarization-community-1 5.3 M whisper-large-v3 4.7 M