Jukebox

Royalty-free music for restaurants adapted to customer mood using generative AI.

Comment 2

The entirety of this project was bootstrapped over the course of the TreeHacks weekend. Thank you so much to the organizers!

Inspiration

Boring music kills. Let's face it: dull music can kill a good vibe, especially in restaurants or bars. Our goal? We wanted to provide restaurants with royalty-free, unique way for generate new music that always adapts to and matches the mood of their customers, elevating their experience by creating complete audial immersion into the atmosphere of the restaurant. Picture this: a dynamic soundtrack that not only enhances the dining experience but also immerses everyone in the restaurant's unique atmosphere.

With a psychology student and a professional musician in our team, we understand the importance of music as more than just background noise, it isn't just an ambiance—it's a mood shifter. The psychologist sees music as a method to create a personalized, almost therapeutic experience for the audience, while our musician clearly knows firsthand the power of setting the right tone.

What it does

On check-in, Jukebox takes customers' polls on their current moods and uses it to generate custom music tracks that perfectly match the energy of all customers in the restaurant. We use a generative music AI (Suno) to provide the perfect music for the vibe . When someone enters a new mood, we keep track of their input and in regular intervals, regenerate and adapt the streamed music to current customers.

How we built it

  • Writing custom selenium code to navigate the AI music generation website and download audio
  • Developing a custom streaming method to fetch and blend the next songs automatically
  • Building a split user/restaurant site to provide an interface for users to vote and restaurants to stream music.
  • Iteratively designing and prototyping a welcoming interface in Figma and integrating front-end.
  • Deploying our app onto Firebase for easy access from any device

Challenges we ran into

  • No straightforward API to download/access AI-generated music. This meant we had to spend a lot of time finding workarounds that didn't sacrifice audio quality.
  • Maintaining a continuous stream of different music segments and smoothly linking those generated songs together.

Accomplishments that we're proud of

  • We developed a custom streaming method to stream limitless server-sided music files to our client side.
  • We implemented a method for live download and upload of AI music
  • Clean, well-designed UI experience for ease of use on customer and restaurant side.
  • Intentional design into security and reliability. In total, we are really proud to have been able to integrate so many parts of system into a deployed solution we can test with and demo like with crowds on demo day!

What we learned

Lesson learned: Streaming is hard! (Watch your head(ers)!) We had to navigate some messy sites in hacky ways to get the job done. Most sites are not built cleanly for selenium interactions. Lastly, security is hard.

What's next for Jukebox

We never settle:

  • Further development in the separation of the API into many restaurants design
  • Expansion into the therapeutic aspect of music generation
  • Further development on audio quality and consistency

Built With

Share this project:

Updates