Inspiration - Having elder relatives of my own who are not the most technologically savy, they suffer from the recent outburst of AI scam calls, and he almost lost all the money he had in his ICICI bank account.
What it does - It's an enterprise tool that will eventually be affiliated with banks and insurance companies to scan calls as they happen and inform the employee whether the voice being used is real or a deepfake made to steal their money.
How we built it - combined github supabace vercel and claude api using the sonnet model to handle user authentication with google sign in, account check to see if it actually exists, and github to deploy the code.
Challenges we ran into - inital audio model wasn't working correctely, inital parameters were too bleak to certify if a human voice wasn't ai.
Accomplishments that we're proud of we troubleshooted the audiomodel that wasn't working and set proper parameters that made it possible to accurately tell if the voice is AI or not.
What we learned - How to develop an authentication AI.
What's next for VauxGuard - Upwards scaling and live detection instead of needing an alternate device.
Log in or sign up for Devpost to join the conversation.