we are developing a rechargeable battery-powered iot device to monitor ambient sounds and identify certain audio events (like a specific bird sound, or the word 'yes'). It needs to determine exactly when a specific event occurred.
the device probably needs to use a microprocessor cascade to run increasingly complex ML models to save power. to avoid constantly running the most expensive model.
ideally, solar panels would be used to allow for constant usage in the field.
tensorflow lite will probably be used to run optimized ML algorithms on the device to process the audio feed.
our current challenge is determining ML model constraints given that the model needs to be run continuously on solar-powered batteries. we need to know the minimum processor power required, so we can develop hardware prototype along with ML models.
if you're interested in working with this project to help us with these items, please describe:
- yourself
- your background
- relevant experience
- why you're interested
we have a very small budget so this project might be ideal for students.
About the recuiterMember since Sep 9, 2017 Yosef
from Hazafon, Israel