Generate. Interpolate. Orchestrate.
Group 3A Members
Laura Ibanez
António Ramires
Seunghun Lee
Satvik Venkatesh
Anja Scheller
Ying Wang
CJ
Presenting & Live Demo
Laura & António
#AIxMUSIC20
Introduction
1940s
Rock Music
EDM
1980s
AI-esthetic
2020?
What’s AI-esthetic?
Key concepts
Existing music
Neural transfer
Source separation
Environmental sounds
AI-esthetic
Motivation
Turning everdays environmental sounds into before unheard soundscapes.
Changing the perception of our environment by breaking its patterns and transforming them into a new hearing experience by using AI technology.
Finding a new way to capture the unique atmosphere of a place, turning noises into instrumental sounds and by that creating an additional level of sensual experience.
Project Steps
Finding the right sources = field recordings from different environments
A day at the beach
Park scene
Construction work
Unmix the field recordings & create 4 separate stems for each
Vocals Drums Bass Other
Using DDSP Timbre Transfer* (pre-trained and own trained models)
vocals -> violin
drums -> double bass
bass -> cello
other -> viola
Learnings: When training the models - quality beats quantity
*) Link: tba
“Recreating” the environment by using Mixer & Crossfader
https://soundscape-ai.netlify.app
Thank you for watching & listening!
#AIxMUSIC20