A crucial aspect for the successful deployment of audio-based models "in-the-wild" is the robustness to the transformations introduced by heterogeneous acquisition conditions. In this work, we propose a method to perform one-shot microphone style transfer. Given only a few seconds of audio recorded by a target device, MicAugment identifies the transformations associated with the input acquisition pipeline and uses the learned transformations to synthesize audio as if it were recorded under the same conditions as the target audio. We show that our method can successfully apply the style transfer to real audio and that it significantly increases model robustness when used as data augmentation in the downstream tasks.

We consider the problem of separating a particular sound source from a single-channel mixture, based on only a short sample of the target source (from the same recording). Using SoundFilter, a wave-to-wave neural network architecture, we can train a model without using any sound class labels. Using a conditioning encoder model which is learned jointly with the source separation network, the trained model can be "configured" to filter arbitrary sound sources, even ones that it has not seen during training. Evaluated on the FSD50k dataset, our model obtains an SI-SDR improvement of 9.6 dB for mixtures of two sounds. When trained on Librispeech, our model achieves an SI-SDR improvement of 14.0 dB when separating one voice from a mixture of two speakers. Moreover, we show that the representation learned by the conditioning encoder clusters acoustically similar sounds together in the embedding space, even though it is trained without using any labels.

We present a method to estimate the fundamental frequency in monophonic audio, often referred to as pitch estimation. In contrast to existing methods, our neural network can be fully trained only on unlabeled data, using self-supervision. A tiny amount of labeled data is needed solely for mapping the network outputs to absolute pitch values. The key to this is the observation that if one creates two examples from one original audio clip by pitch shifting both, the difference between the correct outputs is known, without even knowing the actual pitch value in the original clip. Somewhat surprisingly, this idea combined with an auxiliary reconstruction loss allows training a pitch estimation model. Our results show that our pitch estimation method obtains an accuracy comparable to fully supervised models on monophonic audio, without the need for large labeled datasets. In addition, we are able to train a voicing detection output in the same model, again without using any labels.

We explore self-supervision as a way to learn general purpose audio representations. Specifically, we propose two self-supervised tasks: Audio2Vec, which aims at reconstructing a spectrogram slice from past and future slices and TemporalGap, which estimates the distance between two short audio segments extracted at random from the same audio clip. We evaluate how the representations learned via self-supervision transfer to different downstream tasks, either training a task-specific linear classifier on top of the pretrained embeddings, or fine-tuning a model end-to-end for each downstream task. Our results show that the representations learned with Audio2Vec transfer better than those learned by fully-supervised training on Audioset. In addition, by fine-tuning Audio2Vec representations it is possible to outperform fully-supervised models trained from scratch on each task, when limited data is available, thus improving label efficiency.

We propose a model to estimate the fundamental frequency in monophonic audio, often referred to as pitch estimation. We acknowledge the fact that obtaining ground truth annotations at the required temporal and frequency resolution is a particularly daunting task. Therefore, we propose to adopt a self-supervised learning technique, which is able to estimate pitch without any form of supervision. The key observation is that pitch shift maps to a simple translation when the audio signal is analysed through the lens of the constant-Q transform (CQT). We design a self-supervised task by feeding two shifted slices of the CQT to the same convolutional encoder, and require that the difference in the outputs is proportional to the corresponding difference in pitch. In addition, we introduce a small model head on top of the encoder, which is able to determine the confidence of the pitch estimate, so as to distinguish between voiced and unvoiced audio. Our results show that the proposed method is able to estimate pitch at a level of accuracy comparable to fully supervised models, both on clean and noisy audio samples, although it does not require access to large labeled datasets.

Communication in networks suffers if a link fails. When the links are edges of a tree that has been chosen from an underlying graph of all possible links, a broken link even disconnects the network. Most often, the link is restored rapidly. A good policy to deal with this sort of transient link failures is swap rerouting, where the temporarily broken link is replaced by a single swap link from the underlying graph. A rapid replacement of a broken link by a swap link is only possible if all swap links have been precomputed. The selection of high-quality swap links is essential; it must follow the same objective as the originally chosen communication subnetwork. We are interested in a minimum-diameter tree in a graph with edge weights (so as to minimize the maximum travel time of messages). Hence, each swap link must minimize (among all possible swaps) the diameter of the tree that results from swapping. We propose a distributed algorithm that efficiently computes all of these swap links, and we explain how to route messages across swap edges with a compact routing scheme. Finally, we consider the computation of swap edges in an arbitrary spanning tree, where swap edges are chosen to minimize the time required to adapt routing in case of a failure, and give efficient distributed algorithms for two variants of this problem.

The efficient distributed construction of a maximal independent set (MIS) of a graph is of fundamental importance. We study the problem in the class of Growth-Bounded Graphs, which includes for example the well-known Unit Disk Graphs. In contrast to the fastest (time-optimal) existing approach [11], we assume that no geometric information (e.g., distances in the graph's embedding) is given. Instead, nodes employ randomization for their decisions. Our algorithm computes a MIS in O(log log n • log* n) rounds with very high probability for graphs with bounded growth, where n denotes the number of nodes in the graph. In view of Linial's Ω(log* n) lower bound for computing a MIS in ring networks [12], which was extended to randomized algorithms independently by Naor [18] and Linial [13], our solution is close to optimal.