AIbase
EN
Home
Deeepfake Audio Recognition Ttoo
D

Deeepfake Audio Recognition Ttoo

Developed by Hemg
A deepfake audio detection model fine-tuned based on facebook/wav2vec2-base, achieving 95.45% accuracy on the evaluation set
Audio Classification
Transformers
Open Source License:Apache-2.0#Audio forgery detection#High accuracy (95.45%)#wav2vec2 fine-tuning
Downloads 20
Release Time : 4/1/2024

Model Overview

This model is specifically designed to identify deepfake audio, fine-tuned on the wav2vec2 architecture, effectively distinguishing between real and fake audio

Model Features

High accuracy
Achieves 95.45% accuracy on the test set
Based on wav2vec2 architecture
Uses the proven wav2vec2-base as the base model
Lightweight fine-tuning
Requires only 16 training epochs to achieve high performance

Model Capabilities

Audio classification
Deepfake detection
Voice authenticity verification

Use Cases

Security verification
Telephone fraud prevention
Detects potential AI-synthesized voices in phone calls
Can effectively identify over 95% of fake voices
Audio evidence verification
Verifies the authenticity of recorded evidence
Content moderation
Social media content moderation
Automatically detects 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