YN Beats

Music Editing, and uploading page
Landing Page

Inspiration

In the current setting, music has a high barrier to entry with people who are aspiring to be creative through music, like our teammate, having to go through 1,000+ hours of training in that instrument to add to/create a song.

What it does

Our technology helps leverage the power of AI, Machine Learning and Code to make music creation more accessible to users as a whole. Instead of having to learn multiple instruments and waste hours of your time to play a snippet with that instrument, utilize your voice and hum or sing your tunes to define your melody, drums, harmony, etc.

How we built it

We used Vite + Tailwind CSS for our frontend framework which allows for user interaction with our app, Flask for our Python backend server that contains all our AI, ML and music detection and breakdown math. We used Vercel's inbuilt Blob Storage to store and retrieve audio files and this is hosted directly on Vercel for deployment and internet usage.

Challenges we ran into

For our ML model, we required much larger datasets for detection and training and due to the limited time frame of the hackathon, we only were able to train our model on so much data and thus our output from our algorithm in the backend is not as optimal or expected.

Accomplishments that we're proud of

During our Hackathon, the most proud accomplishments that we had that we were proud of were the ML algorithm, the music note Math Detection algorithm we created, and also having a glimpse of where this product could be taken as well. After hearing the product work, as an MVP, we're proud to have worked on this and explored a new field for us.

What we learned

During the building process, we learned that data is the biggest factor in alot of projects. With effective data methodologies or scraping, all models, regardless of foundation or ML, will proceed to fail.

What's next for YN Beats

For YN (Young Neighbor) Beats, next steps would be increasing training data and improving our model detection accuracy while also improving the math for note detection for the DAW (or music production software) so that we can get smoother, and more real-like output notes.