ATAT Innovation Labs

AT-CS32 · Cyber security

Deepfake Voice Call Detection

Analyses call audio and warns when the voice is likely AI-generated.

Get the cost on WhatsApp

Suits
Sem 7–8
Branches
CSE, ISE, AIML
Type
Software

How it works

The audio is converted into features that describe the voice, such as its spectrogram. A deep learning model trained on genuine and AI-generated voices looks for the small artefacts that synthetic speech leaves behind. You can upload a recording or test with live audio, and the result is given as the likelihood that the voice is fake.

What it does

  • Audio features
  • Detection model
  • Upload or live test

Built with

Python, deep learning

What you get

  • Working software with full source code
  • Steps to install and run it on your laptop
  • Rough report and explanation of the working
  • Live demo on video call before the final payment

Good to know

  • Features are as listed here. Anything extra is quoted separately.
  • Runs on a normal laptop unless we tell you otherwise.
  • Full reports and PPTs are not included.

Similar projects