Abstract
Algorithms forreliable real-time score following in live opera promise a lot ofuseful applications such as automatic subtitles display, orreal-time video cutting in live streaming. Until now, such systemswere based on the strong assumption that an opera performancefollows the structure of the score linearly. However, this israrely the case in practice, because of different opera versionsand directors' cutting choices. In this paper, we propose atwo-level solution to this problem. We introduce areal-time-capable, high-resolution (HR) tracker that can handlejumps or repetitions at specific locations provided to it. We thencombine this with an additional low-resolution (LR) tracker thatcan handle all sorts of mismatches that can occur at any time,with some imprecision, and can re-direct the HR tracker if thelatter is `lost' in the score. We show that the combination of thetwo improves tracking robustness in the presence of strongstructural mismatches.
Original language | English |
---|---|
Title of host publication | Proceedings of the 29th European Signal Processing Conference (EUSIPCO2021), Dublin, Ireland. |
Number of pages | 5 |
Publication status | Published - Aug 2021 |
Fields of science
- 202002 Audiovisual media
- 102 Computer Sciences
- 102001 Artificial intelligence
- 102003 Image processing
- 102015 Information systems
JKU Focus areas
- Digital Transformation