The Neuralacoustics Project: Exploring Deep-Learnin for Lightweight Numerical Modeling Synthesis
Other conference contribution, 2022
Workshop, Session 1.
We present a newly started project called Neuralacoustics, aimed at using embedded AI to reduce the computational requirements of numerical modeling synthesis. To achieve this goal, we propose a novel design pipeline, in which neural networks are first trained off-line to approximate the behavior of target interactive acoustic systems, and then deployed on embedded and mobile platforms as real-time synthesis engines.
We present a newly started project called Neuralacoustics, aimed at using embedded AI to reduce the computational requirements of numerical modeling synthesis. To achieve this goal, we propose a novel design pipeline, in which neural networks are first trained off-line to approximate the behavior of target interactive acoustic systems, and then deployed on embedded and mobile platforms as real-time synthesis engines.
musical acoustics
deep learning
non-linear acoustics
machine learning
Author
Victor Zappi
Northeastern University
Kivanc Tatar
Chalmers, Computer Science and Engineering (Chalmers), Data Science and AI
Embedded AI for NIME: Challenges and Opportunities Workshop at New Interfaces for Musical Expression
The International Conference on New Interfaces for Musical Expression
Auckland, New Zealand,
Auckland, New Zealand,
Areas of Advance
Information and Communication Technology
Subject Categories
Embedded Systems