Hi, I'm Matt 👋

I'm an independent developer working at the intersection of machine learning and audio security.

What I'm building

[Jabberjay](https://github.com/MattyB95/Jabberjay) is a unified Python API and CLI for detecting AI-generated voices and audio deepfakes.

Synthetic voice detection research is fragmented: models are spread across research repos with incompatible interfaces, different dependencies, and inconsistent outputs. Jabberjay addresses that by wrapping seven state-of-the-art model families — ViT, AST, Wav2Vec2, HuBERT, WavLM, RawNet2, and a classical baseline — behind one consistent API.

$ jabberjay suspicious_call.wav
Spoof ❌  (97.8% confidence, model=VIT)

Why your support matters

I maintain Jabberjay in my own time alongside the research it's built on. Sponsorships help me:

• Dedicate more time to adding new models and keeping dependencies up to date
• Cover infrastructure costs, including compute for training and evaluating models and Hugging Face storage
• Keep the project MIT-licensed and free for everyone — researchers, developers, and security teams alike

Even a small monthly contribution makes a real difference. Thank you for considering it. 🙏