-
Homepage
-
Visualization of audio spectrum as image pushed into Convolutional Neural Network
Inspiration
Popular recommendation systems such as youtube and spotify use collaborative filtering to recommend songs and videos. However, these are based on people's actions which might result in inaccurate recommendations. We wanted to try a novel method at classifying and recommending songs to users.
What it does
It is a music web app, and users can upload songs to get recommendations based on song that they uploaded. The recommender system was built using Convolutional Neural Networks! The input being a picture of the audio spectrum.
How I built it
Coded in python to train the neural network with flask as the backend, and the frontend was coded using html, css and node.js
Challenges I ran into
Problem with the connection to the flask server!
Accomplishments that I'm proud of
Designing the website, creating a pipeline of training to prediction
What I learned
We need a better computer to train the neural network :')
What's next for Vibe
Clustering of songs to find sub-genres
Log in or sign up for Devpost to join the conversation.