Voice ID Enrollment
Voice ID Verification
Encrypted Password Storage
Ever got logged out of your favourite social network and forgot the password you set years ago? Ever tried to remember a complex enough password to make sure your data is safe?
If you thought Yes, I did for any of the questions, our Biometric Voice ID Password Manager is just what you need!
We build a safe password storage with the following goals:
- Empower people with memory disabilities, like Alzheimers disease, to be able to securely log into their services
- Make safe and convenient password storage available on low-processing power smart devices (Watches, Bands,...)
- Fill in the gap created by Face ID and Fingerprint sensors, making biometric authentication available in settings where those modalities are note available
- Empower vision impaired people by giving them access to their password storage (Future Extension)
What it does
Our password manager works pretty much like a standard password vault (for example the Apple Keychain), except we do not store a master password to unlock the vault, the password is a part of you - your unique voice. This way, our password storage is not only safe, but also easily accessible, especially for people with disabilities and from smart devices.
How we built it
Our Biometric VoiceID Password Manager incorporates 4 software components:
- An iOS app (extended to Android and WebApp in future work)
- The stdlib logic layer between the iOS frontend, the Machine Learning layer and the database
- The Microsoft Azure Cognitive System, providing machine learning capabilities for the speaker recognition
- The Microsoft Azure database, safely storing the encrypted user information and password data
The iOS app is the frontend of our application, capturing the speech data and calling the authentication service via the stdlib logic layer. Using Microsoft Azures Cognitive Systems, we get reliable speaker authentication, which is returned to the iOS app via stdlib. The password storage itself is realized using stdlib to aggregate and verify the database information and send it to the frontend.
Challenges we ran into
- Train the cognitive speaker recognition service with speech data recorded from the iOS app
- Create the necessary API endpoints, connecting with the Azure Cognitive System and the Azure database
Accomplishments that we're proud of
- A fully working iOS application showcasing the power of biometric speech data to unlock a secure storage
- A complete backend infrastructure to store, retrieve and execute the Azure services
What we learned
- A seamless integration of different languages, platforms and frameworks creates unforeseen compatibility issues
- REST APIs are a great way to create independent and modular backend systems
What's next for Biometric VoiceID - Password Manager
- Extension to serve vision impaired people
- Create an Android application and a Webapp to make the secure password storage accessible for everyone