W

Wav2vec2 Xls R 1b Korean

Developed by anantoj
This model is a Korean automatic speech recognition model fine-tuned on the KRESNIK/ZEROTH_KOREAN - CLEAN dataset based on facebook/wav2vec2-xls-r-1b
Downloads 20
Release Time : 3/2/2022

Model Overview

This is an automatic speech recognition model optimized for Korean, based on the 1-billion-parameter XLS-R architecture and fine-tuned on Korean datasets

Model Features

Large-scale pre-training
Based on the 1-billion-parameter XLS-R architecture with powerful speech feature extraction capabilities
Korean optimization
Specially fine-tuned on Korean datasets to adapt to Korean speech characteristics
Efficient training
Uses mixed-precision training and gradient accumulation techniques to improve training efficiency

Model Capabilities

Korean speech recognition
Speech-to-text
Audio content understanding

Use Cases

Speech transcription
Korean meeting minutes
Automatically convert Korean meeting recordings into text transcripts
Achieved WER of 82.07 and CER of 42.12 on test set
Voice assistant
Used as the speech recognition module for Korean voice assistants
Educational technology
Korean learning applications
Help learners practice Korean pronunciation and listening
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
Š 2025AIbase