using tensorflow, build two different audio detection machine learning algorithm/models:
1. identify any percussive/sound, occurring nearby or far away (gunfire, snare drum, fireworks, hammer, etc)
2. identify only gunshots. this model should be able to distinguish gunshots from other similar sounds (fireworks, etc). should identify both nearby and far away gunfire.
each model should be able to input an audio file, and return the precise timestamps in those audio files that the given audio event occurred. the deliverable will include 2 working models, and detailed instructions on how to test the model in a linux ec2 instance.
later, we will need to convert this model to tensorflow-lite for an iot application. generating the initial model in TF-lite is also an option, and would be ideal.
About the recuiterMember since Sep 8, 2017 Maxwell
from Victoria, Australia