In cross-modal music processing, translation between visual, auditory, and semantic content opens up new possibilities as well as challenges. The construction of such a transformative scheme depends upon a benchmark corpus with a comprehensive data infrastructure. In particular, the assembly of a large-scale cross-modal dataset presents major challenges. In this paper, we present the MOSA (Music mOtion with Semantic Annotation) dataset, which contains high quality 3-D motion capture data, aligned audio recordings, and note-by-note semantic annotations of pitch, beat, phrase, dynamic, articulation, and harmony for 742 professional music performances by 23 professional musicians, comprising more than 30 hours and 570 K notes of data. To our knowledge, this is the largest cross-modal music dataset with note-level annotations to date. To demonstrate the usage of the MOSA dataset, we present several innovative cross-modal music information retrieval (MIR) and musical content generation tasks, including the detection of beats, downbeats, phrases, and expressive contents from audio, video and motion data, and the generation of musicians' body motion from given music audio. The dataset and codes are available alongside this publication (https://github.com/yufenhuang/MOSA-Music-mOtion-and-Semantic-Annotation-dataset).

This article concerns the problem of the trajectory design and base station (BS) association for cellular-connected unmanned aerial vehicles (UAVs). To support safety-critical functions, one primary requirement for UAVs is to maintain reliable cellular connectivity at every time instant during the flight mission. Since the antenna gain of a ground BS (GBS) changes with the position of the UAV, the UAV-GBS association strategy should be jointly considered with the trajectory design, which has not been studied in the prior arts. In this article, we first formulate the problem of joint BS association and trajectory design with the objective of minimizing the mission completion time under a connectivity outage constraint. Then, a deep learning framework is proposed to solve the formulated nonconvex optimization problem in a decoupled manner. For the UAV-GBS association strategy, the signal strength radio map of a given area is constructed, which is used to train a deep neural network (DNN) to approximate the nonlinear mapping from the UAV position to the optimal GBS. To tackle the high complexity due to the coupled decision variables of GBS association and UAV movement, a novel deep reinforcement learning (DRL) approach is developed to learn the optimal trajectory, in which the UAV can learn from its own past good experiences. Our simulation results confirm the superiority of the proposed DRL approach compared to the conventional DRL approaches in terms of the trajectory length. Additionally, it is demonstrated that the nearest association scheme fails to provide reliable cellular connections, whereas our proposed approach can ensure strong connectivity with the GBS during the whole trajectory.