Inspiration
I’ve always enjoyed the feeling where you’re listening to music in the car, and once you stop at a red light or you stop driving, the song ends. I thought that that was a really cool coincidence, and I want to share that.
What it does
Dynamidrive lets you upload your own music files to the app, you can either use an on-device AI model to separate your music into the Drums, Bass, Voice, and other instruments. You can also upload your own separated files. The app then has your location to measure your speed, and based on your speed, the tracks’ files fade in or out. You can also choose certain segments of the song to loop so once you accelerate or slow down, the song enters a new part.
How we built it
The Dynamidrive App is solely built in SwiftUI. My process is that I write the code that I think is correct, then ask Codex to make it better. Sometimes I ask it to add an entirely new feature, then refine it from there. Little by little, the project has turned into what it is now.
Challenges we ran into
One of the most difficult things about the project is probably advertising. Personally, I do not want AI making any promotional material for the app, so either alone or with my friends, we will make it ourselves.
Accomplishments that we're proud of
At the beginning, I was using a 3rd party service to perform the functionality of the Audio Separator, and I wanted to directly integrate that into the app, so I started a backend server that has the AI model, so the users can easily upload their files for processing. I used a Google Cloud VM for this, and it came with 2 really big drawbacks. The performance was incredibly slow, and the server cost $70 a month. It was the performance part that was really annoying, because when I showed my friends the project, the time it took to separate the file was the song length and a half. I asked Codex if it was possible to move the model onto the device itself, and it eventually succeeded and now takes around 6 seconds to process a whole 2 and a half minute song. I also no longer have to pay for the server, so that was the biggest success in the project’s development.
What we learned
The biggest thing I learned is that nobody is going to interact with your product if they don’t know it exists. You have to really put yourself and the product out there so others will be enticed to check it out. I have shown off the project at multiple car shows, because that’s the main target audience, and that has gotten me majority of the testers of the app.
What's next for Dynamidrive
As of right now, the app is currently in beta. I plan on releasing it to the App Store this fall. My friend who is invested in this project as well, is considering helping me or making a port to Android phones. After that, I plan on really listening to the users’ feedback, and adding more and more features to the app that they actually want.
Log in or sign up for Devpost to join the conversation.