We are pleased to announce the launch of the 1st DAFx Parameter Estimation Challenge!
The competition focuses on two tasks:
Participation is open to individuals, academic groups and industry teams; each team may submit up to two distinct proposals per task. We encourage the submission of any sort of estimation algorithm!
The DAFx Challenge is now officially open: the dataset has been released publicly!
The dataset includes the synthetic IRs in numpy format (npz) generated with our modal plate model. These are the files to be used for the identification tasks, and they contain the actual displacement wave. Additionally float32 normalized wav files are included for quick listening, these are peak-amplitude-normalized. The dataset does not include the ground truth, which is kept secret—safely locked away in a crocodile-surrounded castle ;)
Only the parameter ranges are provided, among which parameters have been randomly selected. The ground truth will be used by the Challenge organizers to evaluate your results. You must provide an estimate for each IR, the results will be averaged among all IRs. Please find the dataset under:
https://github.com/LOGUNIVPM/1st-DAFx-Challenge/tree/main/2026-DATASET-STRIPPED
Anyone who wishes to participate in the challenge or that is at least thinking about it, can send their contact email via this form:
1st DAFx Challenge Participation Contact Form
Filling the form is not mandatory to be part of the challenge but allows the organizers to send timely updates if needed. The deadline for sending the results to the organizers is May 31st. Good luck!