Inspiration
This project was inspired by the childhood dream of one day having robots around you that you can interact with, especially ones with the personality of one of my favorite characters in video game history - GLaDOS from Portal
What it does
It can hold voice conversations, use the camera for multimodal interaction, and search the web for updated information.
How we built it
I built it using OpenAI for their GPT 4 multimodal model, Picovoice for detecting the wake-word and converting speech to text, Tavily for retrieving the latest information, and Azure for their TTS solution.
Challenges we ran into
I ran into some problems with the TTS when it came to the degree sign and abbreviation for weather-related units, such as 'K' and 'MPH', where it pronounced them weirdly.
Accomplishments that we're proud of
I am proud of the fact that I succeded in finishing this project, which was rather complicated for me as I am a 1st year student. I am also very happy that most of the people who got to try my robot enjoyed using it.
What we learned
I learned how to use Azure and its innovative platform, improved my Linux skills and had lots of fun (when things actually worked)
What's next for GLaDOS AI Voice Assistant
Implementing the functionality to tell it to stop mid-sentence, fixing the pronunciation of abbreviations, and 3D printing a full body for it.
Built With
- azure
- openai
- opencv
- picovoice
- python
- raspberry-pi
- tavily


Log in or sign up for Devpost to join the conversation.