Abstract
Emotional Machines is an interactive installation that builds affective virtual environments through spoken language. In response to the existing limitations of emotion recognition models incorporating computer vision and electrophysiological activity, whose sources are hindered by a head-mounted display, we propose the adoption of speech emotion recognition (from the audio signal) and semantic sentiment analysis. In detail, we use two machine learning models to predict three main emotional categories from high-level semantic and low-level speech features. Output emotions are mapped to audiovisual representation by an end-To-end process. We use a generative model of chord progressions to transfer speech emotion into music and a synthesized image from the text (transcribed from the user's speech). The generated image is used as the style source in the style-Transfer process onto an equirectangular projection image target selected for each emotional category. The installation is an immersive virtual space encapsulating emotions in spheres disposed into a 3D environment. Thus, users can create new affective representations or interact with other previous encoded instances using joysticks. © 2022 Owner/Author.

Abstract
The extraction of harmonic information from musical audio is fundamental for several music information retrieval tasks. In this paper, we propose novel harmonic audio features based on the perceptually-inspired tonal interval vector space, computed as the Fourier transform of chroma vectors. Our contribution includes mid-level features for musical dissonance, chromaticity, dyadicity, triadicity, diminished quality, diatonicity, and whole-toneness. Moreover, we quantify the perceptual relationship between short- and long-term harmonic structures, tonal dispersion, harmonic changes, and complexity. Beyond the computation on fixed-size windows, we propose a context-sensitive harmonic segmentation approach. We assess the robustness of the new harmonic features in style classification tasks regarding classical music periods and composers. Our results align with, slightly outperforming, existing features and suggest that other musical properties than those in state-of-the-art literature are partially captured. We discuss the features regarding their musical interpretation and compare the different feature groups regarding their effectiveness for discriminating classical music periods and composers. © F. Almeida, G. Bernardes, and C. Weiû.

Abstract
Language is closely related to how we perceive ourselves and signify our reality. In this scope, we created Desiring Machines, an interactive media art project that allows the experience of affective virtual environments adopting speech emotion recognition as the leading input source. Participants can share their emotions by speaking, singing, reciting poetry, or making any vocal sounds to generate virtual environments on the run. Our contribution combines two machine learning models. We propose a long-short term memory and a convolutional neural network to predict four main emotional categories from high-level semantic and low-level paralinguistic acoustic features. Predicted emotions are mapped to audiovisual representations by an end-to-end process encoding emotion in virtual environments. We use a generative model of chord progressions to transfer speech emotion into music based on the tonal interval space. Also, we implement a generative adversarial network to synthesize an image from the transcribed speech-to-text. The generated visuals are used as the style image in the style-transfer process onto an equirectangular projection of a spherical panorama selected for each emotional category. The result is an immersive virtual space encapsulating emotions in spheres disposed into a 3D environment. Users can create new affective representations or interact with other previously encoded instances (This ArtsIT publication is an extended version of the earlier abstract presented at the ACM MM22 [1]). © 2023, ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering.

Abstract
FluidHarmony is an algorithmic method for defining a hierarchical harmonic lexicon in equal temperaments. It utilizes an enharmonic weighted Fourier transform space to represent pitch class set (pcsets) relations. The method ranks pcsets based on user-defined constraints: the importance of interval classes (ICs) and a reference pcset. Evaluation of 5,184 Western musical pieces from the 16th to 20th centuries shows FluidHarmony captures 8% of the corpus's harmony in its top pcsets. This highlights the role of ICs and a reference pcset in regulating harmony in Western tonal music while enabling systematic approaches to define hierarchies and establish metrics beyond 12-TET.

Abstract
Current collaborative augmented reality (AR) systems establish a common localization coordinate frame among users by exchanging and comparing maps comprised of feature points. However, relative positioning through map sharing struggles in dynamic or feature-sparse environments. It also requires that users exchange identical regions of the map, which may not be possible if they are separated by walls or facing different directions. In this paper, we present Cappella11Like its musical inspiration, Cappella utilizes collaboration among agents to forgo the need for instrumentation, an infrastructure-free 6-degrees-of-freedom (6DOF) positioning system for multi-user AR applications that uses motion estimates and range measurements between users to establish an accurate relative coordinate system. Cappella uses visual-inertial odometry (VIO) in conjunction with ultra-wideband (UWB) ranging radios to estimate the relative position of each device in an ad hoc manner. The system leverages a collaborative particle filtering formulation that operates on sporadic messages exchanged between nearby users. Unlike visual landmark sharing approaches, this allows for collaborative AR sessions even if users do not share the same field of view, or if the environment is too dynamic for feature matching to be reliable. We show that not only is it possible to perform collaborative positioning without infrastructure or global coordinates, but that our approach provides nearly the same level of accuracy as fixed infrastructure approaches for AR teaming applications. Cappella consists of an open source UWB firmware and reference mobile phone application that can display the location of team members in real time using mobile AR. We evaluate Cappella across mul-tiple buildings under a wide variety of conditions, including a contiguous 30,000 square foot region spanning multiple floors, and find that it achieves median geometric error in 3D of less than 1 meter.

Abstract
Label noise is omnipresent in the annotations process and has an impact on supervised learning algorithms. This work focuses on the impact of label noise on the performance of learning models by examining the effect of random and class-dependent label noise on a binary classification task: quality assessment for photoplethysmography (PPG). PPG signal is used to detect physiological changes and its quality can have a significant impact on the subsequent tasks, which makes PPG quality assessment a particularly good target for examining the impact of label noise in the field of biomedicine. Random and class-dependent label noise was introduced separately into the training set to emulate the errors associated with fatigue and bias in labeling data samples. We also tested different representations of the PPG, including features defined by domain experts, 1D raw signal and 2D image. Three different classifiers are tested on the noisy training data, including support vector machine (SVM), XGBoost, 1D Resnet and 2D Resnet, which handle three representations, respectively. The results showed that the two deep learning models were more robust than the two traditional machine learning models for both the random and class-dependent label noise. From the representation perspective, the 2D image shows better robustness compared to the 1D raw signal. The logits from three classifiers are also analyzed, the predicted probabilities intend to be more dispersed when more label noise is introduced. From this work, we investigated various factors related to label noise, including representations, label noise type, and data imbalance, which can be a good guidebook for designing more robust methods for label noise in future work.