Inspiration
Team reference code: 4D The application was developed from a combination of ideas including neural style transfer and an improved music recommendation app. We wanted to build an application that could efficiently discover new versions of songs with minimal manual effort for both individuals and industries.
What it does
We built a music style transfer application that modifies the style of the audio file based on the genre you selected. In this application, we focused on two categories of generative AI: image generation and music & audio generation.
How we built it
The image generator (Stable Diffusion) was used to generate a new album cover for personalisation purposes while the music generation model (DeepAFx-ST) was used for the style transfer. For the web development, we built it with React.js and Flask.
Challenges we ran into
Finding resources on music style transfer as it's a new area of development.
Accomplishments that we're proud of
- Successfully developing and running the web application.
What we learned
- Existing music style transfer models.
- How to utilise web development tools.
What's next for Team 36_Music Style Transfer
- Improving the app with a better UI
- Implementing models to produce a better performance.
Log in or sign up for Devpost to join the conversation.