W

Wav2vec2 Large Xls R 300m Spanish Custom

Developed by tomascufaro
This is a speech recognition model fine-tuned on the Common Voice Spanish dataset based on the facebook/wav2vec2-xls-r-300m model, achieving a word error rate of 21.17% on the evaluation set.
Downloads 15
Release Time : 3/2/2022

Model Overview

This model is an optimized automatic speech recognition (ASR) model for Spanish, capable of converting Spanish speech into text.

Model Features

Optimized for Spanish
Specifically fine-tuned on Spanish speech data, improving the accuracy of Spanish recognition.
Based on wav2vec2-xls-r architecture
Utilizes the large-scale self-supervised speech representation learning architecture developed by Facebook.
Relatively lightweight
With 300M parameters, it maintains performance while reducing computational resource requirements.

Model Capabilities

Spanish speech recognition
Speech-to-text
Audio content transcription

Use Cases

Speech transcription
Meeting minutes
Automatically converts Spanish meeting recordings into text transcripts.
Achieves a 21.17% word error rate on the evaluation set.
Voice assistant
Used as a speech recognition component for Spanish voice assistant applications.
Accessibility applications
Real-time caption generation
Generates real-time captions for Spanish video content.
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
Š 2025AIbase