Paper Archive

We release Amp-Space, a large-scale dataset of paired audio samples: a source audio signal, and an output signal, the result of a timbre transformation. The types of transformations we study are from blackbox musical tools (amplifiers, stompboxes, studio effects) traditionally used to shape the sound of guitar, bass, or synthesizer sounds. For each sample of transformed audio, the set of parameters used to create it are given. Samples are from both real and simulated devices, the latter allowing for orders of magnitude greater data than found in comparable datasets. We demonstrate potential use cases of this data by (a) pre-training a conditional WaveNet model on synthetic data and show that it reduces the number of samples necessary to digitally reproduce a real musical device, and (b) training a variational autoencoder to shape a continuous space of timbre transformations for creating new sounds through interpolation.

Download

This paper proposes a novel way of doing audio synthesis at the waveform level using Transformer architectures. We propose a deep neural network for generating waveforms, similar to wavenet . This is fully probabilistic, auto-regressive, and causal, i.e. each sample generated depends on only the previously observed samples. Our approach outperforms a widely used wavenet architecture by up to 9% on a similar dataset for predicting the next step. Using the attention mechanism, we enable the architecture to learn which audio samples are important for the prediction of the future sample. We show how causal transformer generative models can be used for raw waveform synthesis. We also show that this performance can be improved by another 2% by conditioning samples over a wider context. The flexibility of the current model to synthesize audio from latent representations suggests a large number of potential applications. The novel approach of using generative transformer architectures for raw audio synthesis is, however, still far away from generating any meaningful music similar to wavenet, without using latent codes/meta-data to aid the generation process.

Download

In the sinusoidal analysis of sound, using the Short Time Fourier Transform (STFT), there is the assumption that the signal is locally stationary within each FFT frame. If, as in practice, this assumption is violated, the spectrum becomes distorted. Phase Distortion Analysis (PDA) was introduced in 1995 [1] to enhance the analysis of degraded peaks, by using the distortion itself as a source of information about the signal nonstationarity. It was shown that the first order frequency and amplitude modulation could be measured from the degree of phase shift close to the maximum of the mainlobe peak. This paper presents advances with the PDA technique, in particular a neural network implementation that makes estimation robust to noise. The capability to analyse nonstationarities relaxes the restraint on keeping the FFT analysis window short and therefore effectively improves time-frequency resolution. This, in turn, promises greater analysis-synthesis quality through improved identification and tracking of partials during the analysis phase.

Download

Component-wise circuit modeling, also known as “white-box” modeling, is a well established and much discussed technique in virtual analog modeling. This approach is generally limited in accuracy by lack of access to the exact component values present in a real example of the circuit. In this paper we show how this problem can be addressed by implementing the white-box model in a differentiable form, and allowing approximate component values to be learned from raw input–output audio measured from a real device.

Download

Low frequency oscillator (LFO) driven audio effects such as phaser, flanger, and chorus, modify an input signal using time-varying filters and delays, resulting in characteristic sweeping or widening effects. It has been shown that these effects can be modeled using neural networks when conditioned with the ground truth LFO signal. However, in most cases, the LFO signal is not accessible and measurement from the audio signal is nontrivial, hindering the modeling process. To address this, we propose a framework capable of extracting arbitrary LFO signals from processed audio across multiple digital audio effects, parameter settings, and instrument configurations. Since our system imposes no restrictions on the LFO signal shape, we demonstrate its ability to extract quasiperiodic, combined, and distorted modulation signals that are relevant to effect modeling. Furthermore, we show how coupling the extraction model with a simple processing network enables training of end-to-end black-box models of unseen analog or digital LFO-driven audio effects using only dry and wet audio pairs, overcoming the need to access the audio effect or internal LFO signal. We make our code available and provide the trained audio effect models in a real-time VST plugin1 .

Download

Voice conversion has gained increasing popularity within the field of audio manipulation and speech synthesis. Often, the main objective is to transfer the input identity to that of a target speaker without changing its linguistic content. While current work provides high-fidelity solutions they rarely focus on model simplicity, high-sampling rate environments or stream-ability. By incorporating speech representation learning into a generative timbre transfer model, traditionally created for musical purposes, we investigate the realm of voice conversion generated directly in the time domain at high sampling rates. More specifically, we guide the latent space of a baseline model towards linguistically relevant representations and condition it on external speaker information. Through objective and subjective assessments, we demonstrate that the proposed solution can attain levels of naturalness, quality, and intelligibility comparable to those of a state-of-the-art solution for seen speakers, while significantly decreasing inference time. However, despite the presence of target speaker characteristics in the converted output, the actual similarity to unseen speakers remains a challenge.

Download

Extraction of stationary and transient components from audio has many potential applications to audio effects for audio content production. In this paper we explore stationary/transient separation using convolutional autoencoders. We propose two novel unsupervised algorithms for individual and and joint separation. We describe our implementation and show examples. Our results show promise for the use of convolutional autoencoders in the extraction of sparse components from audio spectrograms, particularly using monophonic sounds.

Download

The separation of musical instruments acoustically mixed in one source is a very active field which has been approached from many different viewpoints. This article compares the blind source separation perspective and oscillatory correlation theory taking the auditory scene analysis as the point of departure (ASA). The former technique deals with the separation of a particular signal from a mixture with many others from a statistical point of view. Through the standard Independent Component Analysis (ICA), a blind source separation can be done using the particular and the mixed signals' statistical properties. Thus, the technique is general and does not use previous knowledge about musical instruments. In the second approach, an ASA extension is studied with a dynamic neural model which is able to separate the different musical instruments taking a priori unknown perceptual elements as a point of departure. Applying an inverse transformation to the output of the model, the different contributions to the mixture can be recovered again in the time domain.

Download

Virtual Analog (VA) modeling aims to simulate the behavior of hardware circuits via algorithms to replicate their tone digitally. Dynamic Range Compressor (DRC) is an audio processing module that controls the dynamics of a track by reducing and amplifying the volumes of loud and quiet sounds, which is essential in music production. In recent years, neural-network-based VA modeling has shown great potential in producing high-fidelity models. However, due to the lack of data quantity and diversity, their generalization ability in different parameter settings and input sounds is still limited. To tackle this problem, we present Solid State Bus-Comp, the first large-scale and diverse dataset for modeling the classical VCA compressor — SSL 500 G-Bus. Specifically, we manually collected 175 unmastered songs from the Cambridge Multitrack Library. We recorded the compressed audio in 220 parameter combinations, resulting in an extensive 2528-hour dataset with diverse genres, instruments, tempos, and keys. Moreover, to facilitate the use of our proposed dataset, we conducted benchmark experiments in various open-sourced black-box and grey-box models, as well as white-box plugins. We also conducted ablation studies in different data subsets to illustrate the effectiveness of the improved data diversity and quantity. The dataset and demos are on our project page: https: //www.yichenggu.com/SolidStateBusComp/.

Download

Virtual analog (VA) audio effects are increasingly based on neural networks and deep learning frameworks. Due to the underlying black-box methodology, a successful model will learn to approximate the data it is presented, including potential errors such as latency and audio dropouts as well as non-linear characteristics and frequency-dependent phase shifts produced by the hardware. The latter is of particular interest as the learned phase-response might cause unwanted audible artifacts when the effect is used for creative processing techniques such as dry-wet mixing or parallel compression. To overcome these artifacts we propose differentiable signal processing tools and deep optimization structures for automatically tuning all-pass filters to predict the phase response of different VA simulations, and align processed signals that are out of phase. The approaches are assessed using objective metrics while listening tests evaluate their ability to enhance the quality of parallel path processing techniques. Ultimately, an overparameterized, BiasNet-based, all-pass model is proposed for the optimization problem under consideration, resulting in models that can estimate all-pass filter coefficients to align a dry signal with its affected, wet, equivalent.

Download