Parametric Tuning of Extended Reverberation Algorithm Using Neural Network

Master Thesis by Søren Vøgg K. Lyster

Sound and Music Computing, 2022-05

Feedback Delay Network Reverberator

A custom implementation of a feedback delay network reverberator has been made as a VST3 using JUCE.

The source code to the repository can be found here, and the latest version compiled for x86_64 Ubuntu 18.04 can be found here

---

Neural Network

The neural network for estimating the reverberator parameters can be found here. The Jupyter Notebook ReverberatorEstimator.ipynb contains the setup that has been used to create the network.

---

Evaluation

The evaluation has been done with three different blackbox reverberators. Impulse responses have been created for all three cases with the following input audio:

Hall Of Fame 2 - By TC Electronic

Impulse response (Target / Estimated):

Drums (Blackbox / Estimated):

Piano (Blackbox / Estimated):

Vocals (Blackbox / Estimated):

Figures for blackbox/estimated

Ableton Live Reverb - By Ableton

Impulse response (Target / Estimated):

Drums (Blackbox / Estimated):

Piano (Blackbox / Estimated):

Vocals (Blackbox / Estimated):

Figures for blackbox/estimated

Valhalla Room - By ValhallaDSP

Impulse response (Target / Estimated):

Drums (Blackbox / Estimated):

Piano (Blackbox / Estimated):

Vocals (Blackbox / Estimated):

Figures for blackbox/estimated