Neural networks have the capacity to learn style from an image and apply it to another image. We wanted to investigate the possibility of applying this technique to audio files.
What it does
The user uploads an audio file. They then select a genre of music and 'warp' the uploaded file to make it sound more like that genre.
How we built it
Webhosting services (AWS,GCP) for computation and serving clients. A github project based in Tensor flow for machine learning. Front end development using standard languages (HTML,CSS,JS).
Challenges we ran into
- Cloud hosting services have extremely frustrating security features that make file sharing difficult.
- Django is a powerful, but un-intuitive framework that was hard to learn and implement quickly.
- Setting up Tensor Flow was rather difficult and time consuming
Accomplishments that we're proud of
We managed to apply the idea in a bare-bones form. It is not refined, but the proof of concept exists.
What we learned
Hackathons are very fun and worth anybody's time!
What's next for AudioWarp
Refining the User Experience and optimizing the machine learning parameters to create better sounding results.