Although there is a ton of various applications on the market, only a few apps are installed on every single device and they are messengers. The idea was to try to integrate maximum functionality into one app (messenger) for users not to bother using different applications for particular services.
What it does
The project is a messenger bot that gives access to user's personal banking account based on voice recognition. You can ask common questions about the bank, view your balance and even transfer funds.
How I built it
The voice verification algorithm was built using Gaussian Mixture Models
Challenges I ran into
The duration of a voice message for training the model
Accomplishments that I'm proud of
The verification is based solely on the user's voice regardless of words that they may say. It is easier to use because it prevents from learning an unnecessary piece of text.
What I learned
We learned Telegram API and got closer to specificity of speech-related machine learning algorithms.
What's next for SafeBank
The next step for SafeBank is obtaining semantic information from the user's voice input that is the bot will do whatever you tell him to do. That will make it even easier to get answers and access to banking services.