W

Wav2vec2 Large Xlsr 53 Latvian

Developed by anton-l
This is an automatic speech recognition (ASR) model fine-tuned on the Latvian Common Voice dataset based on Facebook's Wav2Vec2-Large-XLSR-53 model.
Downloads 137
Release Time : 3/2/2022

Model Overview

This model is specifically designed for Latvian speech recognition tasks, fine-tuned on the Common Voice dataset, and supports voice input with a 16kHz sampling rate.

Model Features

High-accuracy Latvian recognition
Achieves 26.89% WER (Word Error Rate) on the Common Voice test set
Based on XLSR pre-trained model
Fine-tuned using the cross-lingual speech representation learning (XLSR) pre-trained model
No language model required
Can be used directly without additional language model support

Model Capabilities

Latvian speech recognition
16kHz audio processing
End-to-end speech-to-text

Use Cases

Speech transcription
Latvian speech-to-text
Convert Latvian speech content into text
26.89% WER
Voice assistants
Latvian voice command recognition
Used for voice command recognition in Latvian voice assistants or control systems
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
Š 2025AIbase