AIbase
EN
Home
Wav2vec2 S F O 8batch 5sec 0.0001lr Unfrozen
W

Wav2vec2 S F O 8batch 5sec 0.0001lr Unfrozen

Developed by reralle
A speech processing model fine-tuned based on facebook/wav2vec2-large, supporting speech recognition tasks
Speech Recognition
Transformers
Open Source License:Apache-2.0#Speech Feature Extraction#Mini-batch Optimization#Short Audio Processing
Downloads 21
Release Time : 5/5/2023

Model Overview

This model is a fine-tuned version based on the facebook/wav2vec2-large architecture, primarily used for speech-related tasks, achieving 66.67% accuracy and 67.42% F1 score on the evaluation set.

Model Features

Efficient Fine-tuning
Fine-tuned based on the pre-trained wav2vec2-large model, fully leveraging the advantages of large-scale pre-training
Optimized Training
Trained with a batch size of 8 and a learning rate of 0.0001, ensuring training stability
Linear Learning Rate Scheduling
Uses a linear learning rate scheduler with a warm-up ratio of 0.003, optimizing the training process

Model Capabilities

Speech Recognition
Audio Feature Extraction

Use Cases

Speech Processing
Speech-to-Text
Convert speech signals into text content
Achieved 66.67% accuracy on the evaluation set
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
English简体中文繁體中文にほんご
© 2025AIbase