Nowadays, it is hard to discover new music; "good music" is confined to whatever that is popular on mainstream media and social media. We wanted a new way to recommend songs to people that preserves their individual tastes and flair.

What it does

This app recommends a song based on elements it detects from a picture uploaded by the user.

How I built it

We made use of Google Vision to identify the key elements of a picture, and then using Spotify's API, generate a song that relates to those elements.

Challenges I ran into

We faced issues with the API calls as there is a small community using these technologies, so there was little information available apart from the documentation provided by the companies themselves, which were not extensive enough.

Accomplishments that I'm proud of

We were able to create something that we genuinely had fun using.

What I learned

It is important to build upon what others have done instead of re-inventing the wheel.

What's next for Picturesque Songs

Incorporating NLP to further extend the recommendation functionality: related words/synonyms/antonyms could be generated so that a more diverse range of songs can be recommended.

Built With

Share this project: