Neural Audio Effect Modelling Strategies for a Spring Reverb
Virtual analog modelling emulates the processing characteristics of a given physical device. This has been an active field of research and commercial innovation in which two main perspectives have been historically adopted. The first one: white-box, seeks to reproduce the exact behaviour through algorithmic simulation of circuits or physical phenomena. The second one: black-box, aims to learn the approximation function from examples recorded at the input and output stages of the target device. In this second approach, deep learning has emerged as a valuable strategy for linear systems, such as filters, as well as nonlinear time-dependent ones like distortion circuits or compressors. The spring reverb is as audio effect with a very long and rooted history in music production and performance, based on a relatively simple design, this device is an effective tool for artificial reverberation. The electromechanical functioning of this reverb makes it a nonlinear timeinvariant spatial system that is difficult to fully emulate in the digital domain with white-box modelling techniques. This thesis wants to address the modelling of spring reverb, leveraging end-to-end neural audio effect architectures through supervised learning. Recurrent, convolutional and hybrid models have successfully been used for similar tasks, especially compressors and distortion circuits emulations. Using two available datasets of guitar recordings, we evaluate with quantitative metrics, acoustical analysis and signal processing measurements the efficiency and the results of four neural network architectures to model this effect. We present the results and outline different strategies for this modelling task, providing a reproducible experimental environment with code.
Code (1)
Similar Papers 제목 키워드 기반
Evaluating Neural Networks Architectures for Spring Reverb Modelling
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 …
Modeling plate and spring reverberation using a DSP-informed deep neural network
Plate and spring reverberators are electromechanical systems first used and researched as means to substitute real room reverberation. Nowadays they are often used in music production for aesthetic reasons due to their p…
Audio-Driven Reinforcement Learning for Head-Orientation in Naturalistic Environments
Although deep reinforcement learning (DRL) approaches in audio signal processing have seen substantial progress in recent years, audio-driven DRL for tasks such as navigation, gaze control and head-orientation control in…
Audio Signal ProcessingDeep Reinforcement LearningQ-Learningreinforcement-learning+1AdVerb: Visually Guided Audio Dereverberation
We present AdVerb, a novel audio-visual dereverberation framework that uses visual cues in addition to the reverberant sound to estimate clean audio. Although audio-only dereverberation is a well-studied problem, our app…
Speaker VerificationSpeech Enhancementspeech-recognitionSpeech RecognitionLearning Audio-Visual Dereverberation
Reverberation not only degrades the quality of speech for human perception, but also severely impacts the accuracy of automatic speech recognition. Prior work attempts to remove reverberation based on the audio modality …
Automatic Speech RecognitionAutomatic Speech Recognition (ASR)Speaker IdentificationSpeech Enhancement+2