6.3 User-generated and multi-perspective concert
video
Finally, it is of relevance to mention recent approaches regardingmulti-
perspective and user-generated concert video
content. This topic has been emerging in several recent
works, and since such content reflects collective strategies
taking into account a particular person’s view on a concert,
it can be of interest for PHENICX too.
As for existing work,in [61] audio fingerprints are used to
synchronise multiple user-generated concert video recordings,
and key moments within a concert are detected based
on the amount of overlap betweenmultiple user clips. In [62],
an automated video mashup system is presented, synchronising
different user videos through camera flashes, and
generating an aesthetic mashup result based on formalised
requirements as elicited from video camera users. Finally,
in [63] a concert video browser is demonstrated based on
segment-level visual concept detectors, in which crowdsourcing
mechanisms are used to improve the indexing results.
It is striking that none of these existing methods actually
base their analyses or evaluations on musical audio
content, nor do they try to relate obtained results to musical
content. In contrast, in PHENICX, since multi-perspective
video and social information are to be used to get a better
insight into the live musical performance, musical aspects
will need to be taken into account explicitly