Deploy TFlite model to ESP32 for sound recognition
Budget: $250 – $750 USD
What i need:
The ESP32 should be able to recognize elephant trumpet sounds.
What i have:
- An ESP32 with a I2S microphone (ICS-43434) and a working firmware that is able to record sound to an SD card in very good quality.
- A already TFlite model from EdgeImpulse
What you should do:
Deploy the existing TFlite model to the existing firmware. The ESP32 should listen continuously and when it detects a elephant trumpet sound it should start recording for 10 seconds and make an LED blink when a trumpet was detected.
And if it's not too much work: Send a https command via Wifi when a sound was detected.
What is the problem?
I only know basics if it comes to programming. So that is over my head :) EdgeImpulse can export a Arduino library to make it work for ESP32. But the library's I2S driver is not compatible with my microphone. So it seems the easiest way is to deploy the C++ library model into my existing firmware which has a working I2S driver for my microphone. More info about that here in this forum post:
https://forum.edgeimpulse.com/t/different-sound-qualities-when-sampling-with-different-frequencies-and-microphones/6614/7
Only the last two post should be relevant.
The C++ library from EdgeImpulse which includes the model is in the attachment.
The existing firmware which records sound can be found here:
https://github.com/EDsteve/idf-wav-sdcard
Would be great to have that within a few days, a week or two weeks?
The ESP32 should be able to recognize elephant trumpet sounds.
What i have:
- An ESP32 with a I2S microphone (ICS-43434) and a working firmware that is able to record sound to an SD card in very good quality.
- A already TFlite model from EdgeImpulse
What you should do:
Deploy the existing TFlite model to the existing firmware. The ESP32 should listen continuously and when it detects a elephant trumpet sound it should start recording for 10 seconds and make an LED blink when a trumpet was detected.
And if it's not too much work: Send a https command via Wifi when a sound was detected.
What is the problem?
I only know basics if it comes to programming. So that is over my head :) EdgeImpulse can export a Arduino library to make it work for ESP32. But the library's I2S driver is not compatible with my microphone. So it seems the easiest way is to deploy the C++ library model into my existing firmware which has a working I2S driver for my microphone. More info about that here in this forum post:
https://forum.edgeimpulse.com/t/different-sound-qualities-when-sampling-with-different-frequencies-and-microphones/6614/7
Only the last two post should be relevant.
The C++ library from EdgeImpulse which includes the model is in the attachment.
The existing firmware which records sound can be found here:
https://github.com/EDsteve/idf-wav-sdcard
Would be great to have that within a few days, a week or two weeks?