We present here the Extended Ballroom dataset. This dataset is an improved version of the well-known Ballroom dataset. It provides amongst other things, more tracks than the Ballroom dataset and a list of track repetitions (exact duplicates , karaoke versions,. . .). Thus it extends the range of possible applications. We describe here how we assembled the Extended Ballroom dataset, how the various annotations (tempo, rhythm class, duplicates, ...) were made and how they can be trusted, as well as the possible applications we can make out of the Extended Ballroom dataset.