Inspiration

By defining Nostalgia 'a sentimental longing or wistful affection for the past, typically for a period or place with happy personal associations', we (former team) believed that by working backwards – we can find a way to trigger these emotions via music. When listening to music, our brains release serotonin (which is what happens when we have a nostalgic feeling), and by realizing that, we targetted trying to trigger those chemicals releases leading a person to get to this 'nostalgic state'. Since nostalgia is subjective and each person feels it differently based on their past and previous associations, we wanted to be able to predict how to curate the perfect nostalgic playlist for users by using data from their past.

What it does

NostalgicLists uses AI and inputs about a human's past to be able to curate a playlist for a user that's meant to be nostalgic when listened to.

How we built it

At first, we tried to use Cohere but then there were some timeout problems that we couldn't control on our side and decided that ChatGPT was more optimal (but we would've used Cohere had it worked at the time). We give ChatGPT the inputs about a user, ask it to generate a list of songs that are most likely to be nostalgic to the user given their past. With that list, we then do a look up on Spotify's side (via API), get the Spotify IDs of all the songs (from Spotify), then create a new playlist and add those songs to it by ID.

Challenges we ran into

Being able to receive the suggested songs as JSON and parse the list into something that's digestible by Spotify's API. We also ran into lag-times between getting suggestions of songs, then doing lookups for each song.

Accomplishments that we're proud of

Being able to talk to Spotify's API super reliably and not have to do any weird work arounds.

What we learned

Figma, React, Next.js, Prisma, REST API calls.

What's next for NostalgicLists

Being able to pull info from third-party oauth providers (Facebook, Google, etc.) to make it even more personalized and accurate. Another feature can be the ability to have AI-generated nostalgic playlist covers.

Built With

Share this project:

Updates