TL;DR

We open-source a PyTorch-based nonlinear string sound synthesizer, which operates on a finite difference time domain.

Paper link

String Sound Synthesizer on GPU-accelerated Finite Difference Scheme

Research Highlights

• The provision of an open-source nonlinear string synthesizer.
• The synthesizer adeptly simulates the vibration of a planar stiff string, accommodating frequency-dependent loss and various excitations, all achieved through FDTD.
• The synthesizer can operate with randomly augmented control parameters and material properties, thus serving as a versatile dataset construction toolkit.
• The synthesizer offers flexibility by supporting both CPU and GPU computations, thereby providing parallelization options.

⚠️ We are still working on this demo page, so stay tuned! ⚠️

Glissando (descending)

Bow

output-u.wav

Hammer

output-u.wav

Pluck

output-u.wav

Glissando (ascending)

output-u.wav

output-u.wav

output-u.wav