Sorry, you need to enable JavaScript to visit this website.

Recent deep learning approaches have shown great improvement in audio source separation tasks. However, the vast majority of such work is focused on improving average separation performance, often neglecting to examine or control the distribution of the results. In this paper, we propose a simple, unified gradient reweighting scheme, with a lightweight modification to bias the learning process of a model and steer it towards a certain distribution of results. More specifically, we reweight the gradient updates of each batch, using a user-specified probability distribution.

The ability of an autonomous vehicle to perform 3D tracking is essential for safe planing and navigation in cluttered environments. The main challenges for multi-object tracking (MOT) in autonomous driving applications reside in the inherent uncertainties regarding the number of objects, when and where the objects may appear and disappear, and uncertainties regarding objects' states. Random finite set (RFS) based approaches can naturally model these uncertainties accurately and elegantly, and they have been widely used in radar-based tracking applications.

The ability of an autonomous vehicle to perform 3D tracking is essential for safe planing and navigation in cluttered environments. The main challenges for multi-object tracking (MOT) in autonomous driving applications reside in the inherent uncertainties regarding the number of objects, when and where the objects may appear and disappear, and uncertainties regarding objects' states. Random finite set (RFS) based approaches can naturally model these uncertainties accurately and elegantly, and they have been widely used in radar-based tracking applications.

Most music source separation systems require large collections of isolated sources for training, which can be difficult to obtain. In this work, we use musical scores, which are comparatively easy to obtain, as a weak label for training a source separation system. In contrast with previous score-informed separation approaches, our system does not require isolated sources, and score is used only as a training target, not required for inference.

Most music source separation systems require large collections of isolated sources for training, which can be difficult to obtain. In this work, we use musical scores, which are comparatively easy to obtain, as a weak label for training a source separation system. In contrast with previous score-informed separation approaches, our system does not require isolated sources, and score is used only as a training target, not required for inference.

The remarkable performance of deep neural networks (DNNs) currently makes them the method of choice for solving linear inverse problems. They have been applied to super-resolve and restore images, as well as to reconstruct MR and CT images. In these applications, DNNs invert a forward operator by finding, via training data, a map between the measurements and the input images. It is then expected that the map is still valid for the test data. This framework, however, introduces measurement inconsistency during testing.

Recent attempts show that factorizing 3D convolutional filters into separate spatial and temporal components brings impressive improvement in action recognition. However, traditional temporal convolution operating along the temporal dimension will aggregate unrelated features, since the feature maps of fast-moving objects have shifted spatial positions. In this paper, we propose a novel and effective Multi-Directional convolution (MDConv), which extracts features along different spatial-temporal orientations.

The paper deals with the hitherto neglected topic of audio dequantization. It reviews the state-of-the-art sparsity-based approaches and proposes several new methods. Convex as well as non-convex approaches are included, and all the presented formulations come in both the synthesis and analysis variants. In the experiments the methods are evaluated using the signal-to-distortion ratio (SDR) and PEMO-Q, a perceptually motivated metric.

CNN for time series data implicitly assumes that the data are uniformly sampled, whereas many event-based and multi-modal data are nonuniform or have heterogeneous sampling rates. Directly applying regular CNN to nonuniform time series is ungrounded, because it is unable to recognize and extract common patterns from the nonuniform input signals. In this paper, we propose the Continuous CNN (\myname), which estimates the inherent continuous inputs by interpolation, and performs continuous convolution on the continuous input.

This work is concerned with non-parallel voice conversion. In particular, motivated by the recent advances in mel-spectrogram-based vocoders, we focus on conversions in the mel-spectrogram domain based on CycleGAN. The challenge is how to make the converter able to convert only the voice factors while retaining the linguistic content factors that underlie input mel-spectrograms. To solve this, we propose MaskCycleGAN-VC, which is an extension of CycleGAN-VC2 and is trained using a novel auxiliary task called filling in frames (FIF). With FIF, we apply a temporal mask to the input mel-spectrogram and encourage the converter to fill in missing frames based on surrounding frames. This task allows the converter to learn time-frequency structures in a self-supervised manner. A subjective evaluation of the naturalness and speaker similarity showed that MaskCycleGAN-VC outperformed previous CycleGAN-VCs.