Hi everyone,
I’m able to run an EI audio recognition model on my rpi model 4b (Locally and on the browser) i want to integrate the model and change it so it sends data to ( twilio / telegram ) when it detects specific sounds, i followed this tutorial GitHub - edgeimpulse/example-linux-with-twilio: Object detection which sends a message over Twilio but it’s for Object recognition and i don’t know how to change it or where to start to make it for audio Classification. Any Help will be much appreciated
Thank you
Hello @Junior.F,
You can have a look at the Audio examples here: linux-sdk-python/examples at master · edgeimpulse/linux-sdk-python · GitHub
This is using the Python SDK but Twilio also provides libraries for Python: The Twilio Python Helper Library | Twilio
If you prefer to stick with NodeJS, here you have two examples:
You’ll need to add something like in the example your provided:
try {
await twilioClient.messages.create({
body: 'Danger! Both human and elephant seen in close proximity!',
to: process.env.TWILIO_TO || '',
from: process.env.TWILIO_FROM || ''
});
}
catch (ex2) {
let ex = <Error>ex2;
console.warn('Failed to send a message via Twilio', ex.message || ex.toString());
}
In the audioClassifier.on('result',...)
block:
audioClassifier.on('result', (ev, timeMs, audioAsPcm) => {
if (!ev.result.classification) return;
// Put your custom Twilio logic here
// print the raw predicted values for this frame
// (turn into string here so the content does not jump around)
let c = ev.result.classification;
for (let k of Object.keys(c)) {
c[k] = c[k].toFixed(4);
}
console.log('classification', timeMs + 'ms.', c);
});
I hope that helps.
Regards,
Louis