This deliverable provides functionality for Task 5.5 of the TROMPA project (Annotation tools). Annotations are important for TROMPA through a wide cross-section of participants and use-cases. For example, music enthusiasts give feedback on how they perceive music and how it makes them feel. Music scholars share notes and insights among themselves to improve their understanding of compositions and performances. Instrumental players annotate scores as personal aide-memoires during rehearsal and share them in pedagogical contexts. Annotations also help us to improve the functionality of automatic systems. Such systems may need annotations to use as ground truth for building automatic classifiers. This deliverable presents the requirements of an annotation tool that can be used in TROMPA to collect annotations from users for a number of the end-user pilots that are being developed in the project. It presents a data model for storing the collected annotations in a way that integrates with the existing tools that have already been built to support the research taking part in TROMPA, for example the Contributor Environment (CE). In this deliverable we focus on annotations of audio, however we have worked closely with other project partners who require the annotation of other modalities (for example, Task 3.4, analysis and annotation of music scores) in order to ensure that our work in the design of the annotation schema can also support annotations for these other types of data. Aspects of the presented tooling are integrated with the Digital Score Edition component (D5.2) to support cross-modal annotations anchored in music score encodings, and with the prototype for music enthusiasts (D6.7) for soliciting feedback from users and storing it as annotations, and using these annotations to train models to recommend new music for a user to listen to.
|Year of Publication
TR-D5.5-Annotation Tools v1