Search

We propose a novel approach for the automatic equalization of individual musical instrument tracks. Our method begins by identifying the instrument present within a source recording in order to choose its corresponding ideal spectrum as a target. Next, the spectral difference between the recording and the target is calculated, and accordingly, an equalizer matching model is used to predict settings for a parametric equalizer. To this end, we build upon a differentiable parametric equalizer matching neural network, demonstrating improvements relative to previously established state-of-the-art. Unlike past approaches, we show how our system naturally allows real-world audio data to be leveraged during the training of our matching model, effectively generating suitably produced training targets in an automated manner mirroring conditions at inference time. Consequently, we illustrate how fine-tuning our matching model on such examples considerably improves parametric equalizer matching performance in real-world scenarios, decreasing mean absolute error by 24% relative to methods relying solely on random parameter sampling techniques as a self-supervised learning strategy. We perform listening tests, and demonstrate that our proposed automatic equalization solution subjectively enhances the tonal characteristics for recordings of common instrument types., Comment: 8 pages, 9 figures. Accepted to the 27th International Conference on Digital Audio Effects (DAFx24)

We investigate how to efficiently set up work groups to boost group productivity, individual satisfaction, and learning. Therefore, we conduct a natural field experiment in a compulsory undergraduate course and study differences between self-selected and randomly assigned groups. We find that self-selected groups perform significantly worse on group assignments. Yet, students in self-selected groups learn more and are more satisfied than those in randomly assigned groups. The effect of allowing students to pick group members dominates the effect of different group compositions in self-selected groups: When controlling for the skill, gender, and home region composition of groups, the differences between self-selected and randomly formed groups persist almost unaltered. The distribution of GitHub commits per group reveals that the better average performance of randomly assigned groups is mainly driven by highly skilled individuals distributed over more groups due to the assignment mechanism. Moreover, these highly skilled individuals contribute more to the group in randomly formed groups. We argue that this mechanism explains why self-selected groups perform worse on the projects but acquire more knowledge than randomly formed groups. These findings are relevant for setting up workgroups in academic, business, and governmental organizations when tasks are not constrained to the skill set of specific individuals.

Many of the commonly used datasets for face recognition development are collected from the internet without proper user consent. Due to the increasing focus on privacy in the social and legal frameworks, the use and distribution of these datasets are being restricted and strongly questioned. These databases, which have a realistically high variability of data per identity, have enabled the success of face recognition models. To build on this success and to align with privacy concerns, synthetic databases, consisting purely of synthetic persons, are increasingly being created and used in the development of face recognition solutions. In this work, we present a three-player generative adversarial network (GAN) framework, namely IDnet, that enables the integration of identity information into the generation process. The third player in our IDnet aims at forcing the generator to learn to generate identity-separable face images. We empirically proved that our IDnet synthetic images are of higher identity discrimination in comparison to the conventional two-player GAN, while maintaining a realistic intra-identity variation. We further studied the identity link between the authentic identities used to train the generator and the generated synthetic identities, showing very low similarities between these identities. We demonstrated the applicability of our IDnet data in training face recognition models by evaluating these models on a wide set of face recognition benchmarks. In comparison to the state-of-the-art works in synthetic-based face recognition, our solution achieved comparable results to a recent rendering-based approach and outperformed all existing GAN-based approaches. The training code and the synthetic face image dataset are publicly available ( https://github.com/fdbtrs/Synthetic-Face-Recognition )., Comment: Accepted at CVPR Workshops

Recent deep face recognition models proposed in the literature utilized large-scale public datasets such as MS-Celeb-1M and VGGFace2 for training very deep neural networks, achieving state-of-the-art performance on mainstream benchmarks. Recently, many of these datasets, e.g., MS-Celeb-1M and VGGFace2, are retracted due to credible privacy and ethical concerns. This motivates this work to propose and investigate the feasibility of using a privacy-friendly synthetically generated face dataset to train face recognition models. Towards this end, we utilize a class-conditional generative adversarial network to generate class-labeled synthetic face images, namely SFace. To address the privacy aspect of using such data to train a face recognition model, we provide extensive evaluation experiments on the identity relation between the synthetic dataset and the original authentic dataset used to train the generative model. Our reported evaluation proved that associating an identity of the authentic dataset to one with the same class label in the synthetic dataset is hardly possible. We also propose to train face recognition on our privacy-friendly dataset, SFace, using three different learning strategies, multi-class classification, label-free knowledge transfer, and combined learning of multi-class classification and knowledge transfer. The reported evaluation results on five authentic face benchmarks demonstrated that the privacy-friendly synthetic dataset has high potential to be used for training face recognition models, achieving, for example, a verification accuracy of 91.87\% on LFW using multi-class classification and 99.13\% using the combined learning strategy.

The process of optimizing the latency of DNN operators with ML models and hardware-in-the-loop, called auto-tuning, has established itself as a pervasive method for the deployment of neural networks. From a search space of loop-optimizations, the candidate providing the best performance has to be selected. Performance of individual configurations is evaluated through hardware measurements. The combinatorial explosion of possible configurations, together with the cost of hardware evaluation makes exhaustive explorations of the search space infeasible in practice. Machine Learning methods, like random forests or reinforcement learning are used to aid in the selection of candidates for hardware evaluation. For general purpose hardware like x86 and GPGPU architectures impressive performance gains can be achieved, compared to hand-optimized libraries like cuDNN. The method is also useful in the space of hardware accelerators with less wide-spread adoption, where a high-performance library is not always available. However, hardware accelerators are often less flexible with respect to their programming which leads to operator configurations not executable on the hardware target. This work evaluates how these invalid configurations affect the auto-tuning process and its underlying performance prediction model for the VTA hardware. From these results, a validity-driven initialization method for AutoTVM is developed, only requiring 41.6% of the necessary hardware measurements to find the best solution, while improving search robustness.

Outdoor shooting ranges are subject to noise regulations from local and national authorities. Restrictions found in these regulations may include limits on times of activities, the overall number of noise events, as well as limits on number of events depending on the class of noise or activity. A noise monitoring system may be used to track overall sound levels, but rarely provide the ability to detect activity or count the number of events, required to compare directly with such regulations. This work investigates the feasibility and performance of an automatic detection system to count noise events. An empirical evaluation was done by collecting data at a newly constructed shooting range and training facility. The data includes tests of multiple weapon configurations from small firearms to high caliber rifles and explosives, at multiple source positions, and collected on multiple different days. Several alternative machine learning models are tested, using as inputs time-series of standard acoustic indicators such as A-weighted sound levels and 1/3 octave spectrogram, and classifiers such as Logistic Regression and Convolutional Neural Networks. Performance for the various alternatives are reported in terms of the False Positive Rate and False Negative Rate. The detection performance was found to be satisfactory for use in automatic logging of time-periods with training activity., Comment: Accepted at 27th International Congress of Sound and Vibration (ICSV27)

The success of Deep Artificial Neural Networks (DNNs) in many domains created a rich body of research concerned with hardware accelerators for compute-intensive DNN operators. However, implementing such operators efficiently with complex hardware intrinsics such as matrix multiply is a task not yet automated gracefully. Solving this task often requires joint program and data layout transformations. First solutions to this problem have been proposed, such as TVM, UNIT or ISAMIR, which work on a loop-level representation of operators and specify data layout and possible program transformations before the embedding into the operator is performed. This top-down approach creates a tension between exploration range and search space complexity, especially when also exploring data layout transformations such as im2col, channel packing or padding. In this work, we propose a new approach to this problem. We created a bottom-up method that allows the joint transformation of both compuation and data layout based on the found embedding. By formulating the embedding as a constraint satisfaction problem over the scalar dataflow, every possible embedding solution is contained in the search space. Adding additional constraints and optmization targets to the solver generates the subset of preferable solutions. An evaluation using the VTA hardware accelerator with the Baidu DeepBench inference benchmark shows that our approach can automatically generate code competitive to reference implementations. Further, we show that dynamically determining the data layout based on intrinsic and workload is beneficial for hardware utilization and performance. In cases where the reference implementation has low hardware utilization due to its fixed deployment strategy, we achieve a geomean speedup of up to x2.813, while individual operators can improve as much as x170., Comment: 25 Pages