[AudioSet](https://research.google.com/audioset/) is a large-scale dataset of manually annotated audio events, collected by Google on Youtube. It is a library with 2,084,320 sound clips 10 seconds longs, all labeled by real humans. Music and musical instrument is largely represented, making this dataset useful for music recognition or [sound classification](Sound%20classification.md). Like it is said in the [dataset index](https://research.google.com/audioset/dataset/index.html), AudioSet contains many music genre annotations that could be use in order to train #machine-learning algorithms for music genre recognition. This is the kind of algorithm used in our [IoT](IoT.md) device [OpenLIVE](https://www.openlive.co/), used to make music recording in our studios more accessible. This system allows musicians to take a recording option during their #rehearsal. The recording starts and stop automaticaly at the beginning and end of their booking and their music is automaticaly mixed, mastered and uploaded on the cloud so they access to it. Music genre recognition intervenes on the mixing process. By knowing the music genre, the system can decide in his own which pre-define mixing preset to apply on the track, instead of applying the same mixing and mastering process on every track.