AIbase
EN
Home
Wavlm Base 960h Asv19 Deepfake
W

Wavlm Base 960h Asv19 Deepfake

Developed by abhishtagatya
A deepfake audio detection model fine-tuned based on Microsoft's WavLM-base, achieving excellent performance on the ASVspoof 2019 dataset with an accuracy of 99.79%
Audio Classification
Transformers
#Audio Forgery Detection#High-precision Classification#Voice Security Verification
Downloads 16
Release Time : 3/14/2024

Model Overview

This model is specifically designed for detecting deepfake audio and audio spoofing attacks, suitable for security verification and content authenticity detection scenarios

Model Features

High-precision Detection
Achieves 99.79% accuracy on the evaluation set, effectively identifying deepfake audio
Low Error Rate
Equal Error Rate (EER) is only 0.80%, demonstrating reliable performance in security-sensitive scenarios
Based on WavLM Architecture
Utilizes WavLM's powerful audio feature extraction capabilities, making it particularly suitable for speech-related tasks

Model Capabilities

Audio Classification
Deepfake Detection
Audio Spoofing Identification
Voice Authenticity Verification

Use Cases

Security Verification
Telephone Banking Fraud Detection
Detects deepfake attacks during voice verification processes
Can identify over 99% of fake voice samples
Content Moderation
Social Media Audio Moderation
Identifies fake audio content on platforms
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