Inspiration
We want Roboy to be able to recognize his environment and to interact with it.
What it does
The algorithm classifies sounds in 10 categories like dog barks, street music, children's play etc. Roboy outputs the result via his speech module.
How we built it
- Python
- Scikit Learn
- Urban Sound Classification: http://aqibsaeed.github.io/2016-09-03-urban-sound-classification-part-1/
- Lots of googling
Challenges we ran into
- Problems with the ROS service
- Feature extraction
- Data preprocessing and data conversion stuff
Accomplishments that we're proud of
- With an accuracy of up to 80% we got multiple machine learning algorithms to perform sound classification
- We used kNNs, Desicion Trees, Random Forests, SVM
- A hopefully real-time sound processing with microphone
What we learned
- Hackathons are fun!
- Scikit Learn is easy to use and very powerful
What's next for Detective Soundboy
We want: unsupervised learning using all the youtube data
Log in or sign up for Devpost to join the conversation.