wav2vec2-large-xlsr-53-hungarian

A Hungarian automatic speech recognition model fine-tuned by Jonatas Grosman from Meta's XLSR-53 multilingual wav2vec2 backbone on Common Voice 6.1 and CSS10 Hungarian data. It converts spoken Hungarian audio to text and reports 31.40% WER and 6.20% CER on the Common Voice Hungarian test set. Source: supabase.