Can I upload two models to the syntiant processor in Arduino Nicla Voice?

Hello, I am working on two models: fall detection using accelerometer values ​​and detection of a word.

Word detection (top)
Fall detection (below)

Both models were implemented using the Syntiant processing block, each model was uploaded to Nicla Voice independently and they work correctly, for each model I generated a binary containing its complete boost.

Now I want both models to upload on nicla voice at the same time, because I want to make the nicla voice send a notification via bluetooth to an app when the user falls or says the preset word.

Is it possible to do this? How could it be done? And if it cannot be done, what recommendations would you give me? Thank you.

Hi @pgudino,

At this time, multiple models/impulses are not supported (outside of separate Linux processes). It’s something we’re looking to support in the future.

Hi, thanks for your response. I have read an article from Edge Impulse that talks about Sensor Fusion. I was trying to merge the inertial and microphone sensors on my Arduino Nicla Voice, but I was not successful. I wanted to ask if it is possible to perform sensor fusion on the Arduino Nicla Voice using the Syntiant processing block in Edge Impulse, or if the Edge Impulse platform does not support this board for sensor fusion. I appreciate your attention and response in advance. Additionally, any suggestions you might have are welcome.

Hi @pgudino,

The issue is that Edge Impulse does not support two different inputs with different sampling rates. Generally, inertial and audio data have different sampling rates. If you can guarantee that the sampling rates are the same, then you can work with a fusion of the data.