AIbase
EN
Home
Wav2vec2 Large Xls R 300m Slovenian
W

Wav2vec2 Large Xls R 300m Slovenian

Developed by bekirbakar
This model is a speech recognition model fine-tuned on the Common Voice Slovenian dataset based on facebook/wav2vec2-xls-r-300m, with a word error rate of 0.3271.
Speech Recognition
Transformers
Open Source License:Apache-2.0#Slovenian speech recognition#High-accuracy speech-to-text#Multilingual pre-training fine-tuning
Downloads 278
Release Time : 6/6/2022

Model Overview

A speech recognition model optimized for Slovenian, fine-tuned on the wav2vec2-xls-r-300m architecture, suitable for speech-to-text tasks.

Model Features

High-performance speech recognition
Achieved a word error rate of 0.3271 on the Common Voice Slovenian dataset
Fine-tuned on a large model
Fine-tuned on the 300-million-parameter wav2vec2-xls-r-300m model, inheriting the powerful feature extraction capabilities of the original model
Optimized training process
Used linear learning rate scheduling with 500 warmup steps, trained for 20 epochs to achieve optimal results

Model Capabilities

Slovenian speech recognition
Audio-to-text conversion
Speech content analysis

Use Cases

Speech transcription
Automated meeting minutes
Automatically convert Slovenian meeting recordings into text transcripts
Accuracy approximately 67.29%
Voice assistant
Provide voice interaction support for Slovenian-speaking users
Educational technology
Language learning applications
Help learners practice Slovenian pronunciation and listening
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
EnglishįŽ€äŊ“中文įšéĢ”中文ãĢãģんご
Š 2025AIbase