Is there a (simple) way to train a node to recognize a specific sound?

After 3 years hearing the neighbor's dog many hundred times a days, it is time to find some help to stop this before I get a nervous breakdown. I want to count the exact times the dog barks and log when he barks, so my question is how to make a solution where node-red can learn to recognize the sound and detect it with good accuracy. I'm thinking now about tensorflow-light or using a spectrum analyzer FFT to detect if the harmonics of the dog barking are present in the correctratio. Any advice how to start? Thanks in advance.

Not sure how well this copes with audio - but it may help - node-red-contrib-teachable-machine (node) - Node-RED
It uses googles teachable machine models which are "easy" to train from samples locally - Teachable Machine
Let us know if it helps/works.

Thank you, will try it out.

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.