What’s all the Fuss about Free Universal Sound Separation Data?
By Scott Wisdom, Hakan Erdogan, D. Ellis, Romain Serizel, Nicolas Turpault, Eduardo Fonseca, Justin Salamon, Prem Seetharaman, John R. Hershey. 2020
96 citations recorded by the source corpus.
We introduce the Free Universal Sound Separation (FUSS) dataset, a new corpus for experiments in separating mixtures of an unknown number of sounds from an open domain of sound types. The dataset consists of 23 hours of single-source audio data drawn from 357 classes, which are used to create mixtures of one to four sources. To simulate reverberation, an acoustic room simulator is used to generate impulse responses of box-shaped rooms with frequency-dependent reflective walls. Additional open-source data augmentation tools are also provided to produce new mixtures with different combinations of sources and room simulations. Finally, we introduce an open-source baseline separation model, based on an improved time-domain convolutional network (TDCN++), that can separate a variable number of sources in a mixture. This model achieves 9.8 dB of scale-invariant signal-to-noise ratio improvement (SI-SNRi) on mixtures with two to four sources, while reconstructing single-source inputs with 35.8 dB absolute SI-SNR. We hope this dataset will lower the barrier to new research and allow for fast iteration and application of novel techniques from other machine learning domains to the sound separation challenge.
Subfields: Cognitive Neuroscience, Signal Processing
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Cognitive Neuroscience
- John R. Hershey
- Scott Wisdom
- Hakan Erdogan
- Marco Tagliasacchi
- Prem Seetharaman
- Eduardo Fonseca
- Justin Salamon
- D. Ellis
In Signal Processing
- Eduardo Fonseca
- Xavier Serra
- Frederic Font
- Xavier Favory
- Jordi Pons
- D. Ellis
- Manoj Plakal
- John R. Hershey
Citations in the corpus
- FSD50K: An Open Dataset of Human-Labeled Sound Events (2020, methodology): “Further, the PP/PNP distinction can be useful for source separation studies [63].”
- Self-Supervised Learning from Automatically Separated Sound Scenes (2021, background): “MixIT does not require knowledge of ground truth source signals, which allows leveraging large amounts of unlabeled data, and has shown great promise in universal sound separation [20, 21].”
- Improving Sound Event Detection in Domestic Environments using Sound Separation (2020, methodology): “The Free Universal Sound Separation (FUSS)(2) dataset [21] is intended for experimenting with universal sound separation [9], and is used as training data for the SS system.”