Examples of build in browser?


I used the platform to generate a model, and loved how intuitive it all was - and the hot word detection worked a treat with the test in the browser, and I can’t wait to use it.

However, once I downloaded the wasm package, I was a little confused as to replicate the microphone/audio piece - as the packaged html file seemed to be just used for classifying the data.

I attempted to build my own audio listener, which would pass the data through to the classifier, but I got extremely different results from the edgeimpulse classifier. I can see the repo for the classifier is open source and listed at the bottom, and I found the Microphone class, so I can possibly extract that out. I just wondered if that would be the correct way to do it, and how I should collate that data to pass through to classify (or classifyContinuous, I’m not sure which to use).

Going to play with it some more to see if I can get to the bottom of it, but thought it might be worth asking here first just in case I’m going down a blind/wrong path :slight_smile: Any help anyone can give would be tremendous. Thank you!

To integrate your own audio listener and pass the data through the Edge Impulse classifier, you’re on the right track. The Edge Impulse platform typically uses TensorFlow Lite for deployment, and you can find the inference code in their GitHub repository.

1 Like