# Differentiable Artificial Reverberation

arXiv: https://arxiv.org/abs/2105.13940 · HTML: https://mlsanity.com/paper/2105.13940/

## Key facts

- Cited by 11 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 28 May 2021.

## Built on by (most cited first, 11 of 11)

- [Music Mixing Style Transfer: A Contrastive Learning Approach to Disentangle Audio Effects](https://mlsanity.com/paper/2211.02247/) (2022)
- [Differentiable Wavetable Synthesis](https://mlsanity.com/paper/2111.10003/) (2021)
- [Adoption of AI Technology in the Music Mixing Workflow: An Investigation](https://mlsanity.com/paper/2304.03407/) (2023)
- [Differentiable Modal Synthesis for Physical Modeling of Planar String Sound and Motion Simulation](https://mlsanity.com/paper/2407.05516/) (2024)
- [Neural Fourier Shift for Binaural Speech Rendering](https://mlsanity.com/paper/2211.00878/) (2022)
- [NaturalL2S: End-to-End High-quality Multispeaker Lip-to-Speech Synthesis with Differential Digital Signal Processing](https://mlsanity.com/paper/2502.12002/) (2025)
- [Data-Driven Room Acoustic Modeling Via Differentiable Feedback Delay Networks With Learnable Delay Lines](https://mlsanity.com/paper/2404.00082/) (2024)
- [Blind Spatial Impulse Response Generation from Separate Room- and Scene-Specific Information](https://mlsanity.com/paper/2409.14971/) (2024)
- [Zero-Shot Mono-to-Binaural Speech Synthesis](https://mlsanity.com/paper/2412.08356/) (2024)
- [Gaussian Process Regression of Steering Vectors With Physics-Aware Deep Composite Kernels for Augmented Listening](https://mlsanity.com/paper/2509.02571/) (2025)
- [Gradient-based Optimisation of Modulation Effects](https://mlsanity.com/paper/2601.04867/) (2026)

## Abstract (arXiv)

Artificial reverberation (AR) models play a central role in various audio applications. Therefore, estimating the AR model parameters (ARPs) of a reference reverberation is a crucial task. Although a few recent deep-learning-based approaches have shown promising performance, their non-end-to-end training scheme prevents them from fully exploiting the potential of deep neural networks. This motivates the introduction of differentiable artificial reverberation (DAR) models, allowing loss gradients to be back-propagated end-to-end. However, implementing the AR models with their difference equations "as is" in the deep learning framework severely bottlenecks the training speed when executed with a parallel processor like GPU due to their infinite impulse response (IIR) components. We tackle this problem by replacing the IIR filters with finite impulse response (FIR) approximations with the frequency-sampling method. Using this technique, we implement three DAR models -- differentiable Filtered Velvet Noise (FVN), Advanced Filtered Velvet Noise (AFVN), and Delay Network (DN). For each AR model, we train its ARP estimation networks for analysis-synthesis (RIR-to-ARP) and blind estimation (reverberant-speech-to-ARP) task in an end-to-end manner with its DAR model counterpart. Experiment results show that the proposed method achieves consistent performance improvement over the non-end-to-end approaches in both objective metrics and subjective listening test results.