Paper Archive

Recent research in deep learning has shown that neural networks can learn differential equations governing dynamical systems. In this paper, we adapt this concept to Virtual Analog (VA) modeling to learn the ordinary differential equations (ODEs) governing the first-order and the second-order diode clipper. The proposed models achieve performance comparable to state-of-the-art recurrent neural networks (RNNs) albeit using fewer parameters. We show that this approach does not require oversampling and allows to increase the sampling rate after the training has completed, which results in increased accuracy. Using a sophisticated numerical solver allows to increase the accuracy at the cost of slower processing. ODEs learned this way do not require closed forms but are still physically interpretable.

Download

Reverberation is a key element in spatial audio perception, historically achieved with the use of analogue devices, such as plate and spring reverb, and in the last decades with digital signal processing techniques that have allowed different approaches for Virtual Analogue Modelling (VAM). The electromechanical functioning of the spring reverb makes it a nonlinear system that is difficult to fully emulate in the digital domain with white-box modelling techniques. In this study, we compare five different neural network architectures, including convolutional and recurrent models, to assess their effectiveness in replicating the characteristics of this audio effect. The evaluation is conducted on two datasets at sampling rates of 16 kHz and 48 kHz. This paper specifically focuses on neural audio architectures that offer parametric control, aiming to advance the boundaries of current black-box modelling techniques in the domain of spring reverberation.

Download

We present a Neural-Driven Multi-Band Processor (NDMP), a differentiable audio processing framework that augments a static sixband Parametric Equalizer (PEQ) with per-band dynamic range compression. We optimize this processor using neural inference for two tasks: Automatic Equalization (AutoEQ), which estimates tonal and dynamic corrections without a reference, and Production Style Transfer (NDMP-ST), which adapts the processing of an input signal to match the tonal and dynamic characteristics of a reference. We train NDMP using a self-supervised strategy, where the model learns to recover a clean signal from inputs degraded with randomly sampled NDMP parameters and gain adjustments. This setup eliminates the need for paired input–target data and enables end-to-end training with audio-domain loss functions. In the inference, AutoEQ enhances previously unseen inputs in a blind setting, while NDMP-ST performs style transfer by predicting taskspecific processing parameters. We evaluate our approach on the MUSDB18 dataset using both objective metrics (e.g., SI-SDR, PESQ, STFT loss) and a listening test. Our results show that NDMP consistently outperforms traditional PEQ and a PEQ+DRC (single-band) baseline, offering a robust neural framework for audio enhancement that combines learned spectral and dynamic control.

Download

Audio processors whose parameters are modified periodically over time are often referred as time-varying or modulation based audio effects. Most existing methods for modeling these type of effect units are often optimized to a very specific circuit and cannot be efficiently generalized to other time-varying effects. Based on convolutional and recurrent neural networks, we propose a deep learning architecture for generic black-box modeling of audio processors with long-term memory. We explore the capabilities of deep neural networks to learn such long temporal dependencies and we show the network modeling various linear and nonlinear, time-varying and time-invariant audio effects. In order to measure the performance of the model, we propose an objective metric based on the psychoacoustics of modulation frequency perception. We also analyze what the model is actually learning and how the given task is accomplished.

Download

Neural audio synthesis methods can achieve high-fidelity and realistic sound generation by utilizing deep generative models. Such models typically rely on external labels which are often discrete as conditioning information to achieve guided sound generation. However, it remains difficult to control the subtle changes in sounds without appropriate and descriptive labels, especially given a limited dataset. This paper proposes an implicit conditioning method for neural audio synthesis using generative adversarial networks that allows for interpretable control of the acoustic features of synthesized sounds. Our technique creates a continuous conditioning space that enables timbre manipulation without relying on explicit labels. We further introduce an evaluation metric to explore controllability and demonstrate that our approach is effective in enabling a degree of controlled variation of different synthesized sound effects for in-domain and cross-domain sounds.

Download

We present two new beat tracking algorithms based on the autocorrelation analysis, which showed state-of-the-art performance in the MIREX 2010 beat tracking contest. Unlike the traditional approach of processing a list of onsets, we propose to use a bidirectional Long Short-Term Memory recurrent neural network to perform a frame by frame beat classification of the signal. As inputs to the network the spectral features of the audio signal and their relative differences are used. The network transforms the signal directly into a beat activation function. An autocorrelation function is then used to determine the predominant tempo to eliminate the erroneously detected - or complement the missing - beats. The first algorithm is tuned for music with constant tempo, whereas the second algorithm is further capable to follow changes in tempo and time signature.

Download

The following article introduces a new parametric synthesis algorithm for sound textures inspired by existing methods used for visual textures. Using a 2D Convolutional Neural Network (CNN), a sound signal is modified until the temporal cross-correlations of the feature maps of its log-spectrogram resemble those of a target texture. We show that the resulting synthesized sound signal is both different from the original and of high quality, while being able to reproduce singular events appearing in the original. This process is performed in the time domain, discarding the harmful phase recovery step which usually concludes synthesis performed in the time-frequency domain. It is also straightforward and flexible, as it does not require any fine tuning between several losses when synthesizing diverse sound textures. Synthesized spectrograms and sound signals are showcased, and a way of extending the synthesis in order to produce a sound of any length is also presented. We also discuss the choice of CNN, border effects in our synthesized signals and possible ways of modifying the algorithm in order to improve its current long computation time.

Download

Deep learning models are mostly used in an offline inference fashion. However, this strongly limits the use of these models inside audio generation setups, as most creative workflows are based on real-time digital signal processing. Although approaches based on recurrent networks can be naturally adapted to this buffer-based computation, the use of convolutions still poses some serious challenges. To tackle this issue, the use of causal streaming convolutions have been proposed. However, this requires specific complexified training and can impact the resulting audio quality. In this paper, we introduce a new method allowing to produce non-causal streaming models. This allows to make any convolutional model compatible with real-time buffer-based processing. As our method is based on a post-training reconfiguration of the model, we show that it is able to transform models trained without causal constraints into streaming models. We apply our method on the recent RAVE model as a case study. This model provides high-quality real-time audio synthesis on a wide range of signals and thus is an ideal candidate to evaluate our method. It should be noted that our method is not restricted to RAVE, and can be straightforwardly applied to any convolutional network. We test our approach on multiple music and speech datasets and show that it is faster than overlap-add methods, while having no impact on the generation quality. Finally, we introduce two open-source implementation of our work as Max/MSP and PureData externals, and as a VST audio plugin. This allows to endow traditional digital audio workstations with real-time neural audio synthesis.

Download

Deep learning models have seen widespread use in modelling LFOdriven audio effects, such as phaser and flanger. Although existing neural architectures exhibit high-quality emulation of individual effects, they do not possess the capability to manipulate the output via control parameters. To address this issue, we introduce Controllable Neural Frame-based Modulation Effects (CONMOD), a single black-box model which emulates various LFOdriven effects in a frame-wise manner, offering control over LFO frequency and feedback parameters. Additionally, the model is capable of learning the continuous embedding space of two distinct phaser effects, enabling us to steer between effects and achieve creative outputs. Our model outperforms previous work while possessing both controllability and universality, presenting opportunities to enhance creativity in modern LFO-driven audio effects. Additional demo of our model is available in the accompanying website.1

Download

This paper investigates the synthesis of audio textures and the structure of generative latent spaces using Variational Autoencoders (VAEs) within two paradigms of neural audio synthesis: DSP-inspired and data-driven approaches. For each paradigm, we propose VAE-based frameworks that allow fine-grained temporal control. We introduce datasets across three categories of environmental sounds to support our investigations. We evaluate and compare the models’ reconstruction performance using objective metrics, and investigate their generative capabilities and latent space structure through latent space interpolations.

Download