W

Wav2vec2 Xls R Parlaspeech Hr

Developed by classla
A Croatian automatic speech recognition model fine-tuned from facebook/wav2vec2-xls-r-300m, trained on 300 hours of parliamentary speech data
Downloads 46.84k
Release Time : 3/2/2022

Model Overview

This model is specifically designed for Croatian automatic speech recognition (ASR) tasks, optimized for parliamentary speech scenarios, capable of accurately transcribing Croatian parliamentary speech content.

Model Features

Parliamentary Speech Optimization
Specially optimized for Croatian parliamentary speech scenarios, capable of handling parliamentary-specific language styles and terminology
High Accuracy
Achieves a character error rate (CER) of 0.0234 and word error rate (WER) of 0.0761 on test sets
Large-scale Training Data
Fine-tuned using 300 hours of Croatian parliamentary speech dataset (ParlaSpeech-HR v1.0)

Model Capabilities

Croatian speech recognition
Parliamentary speech transcription
Audio to text conversion

Use Cases

Government & Parliament
Parliament Meeting Minutes
Automatically transcribe Croatian parliamentary meeting content to generate text records
Accuracy up to 97.66% (based on CER metric)
Academic Research
Political Language Analysis
Provide political science researchers with text data of Croatian parliamentary speeches
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
© 2025AIbase