Running machine learning (ML) models on microcontrollers is one of the most exciting developments of the past years, allowing small battery-powered devices to detect complex motions, recognize sounds, or find anomalies in sensor data. To allow any developer to build these tiny ML models we've partnered with Seeed Studios to bring support for their Wio Terminal to Edge Impulse.
Please prepare wake word or numerical digit identification demonstration with the built-in microphone on the Wio Terminal. Any other examples you can prepare for the Wio Terminal not requiring Grove Sensors
would be appreciated as in the I/O and display sense as well as cost it is the best getting started developer board for Edge Impulse.
What you need is easy on TernsorFlow.There are lots of tutorials.
https://wiki.seeedstudio.com/Wio-Terminal-TensorFlow/
@laefsky and @baozhu1 - apologies for the thread necro, but I’m working on something similar. From what I can tell, the internal mic on the Wio Terminal is an analog mic connected to one of the ADC pins. The only example given is of sampling the ADC as part of the super loop. While you could put that in an interrupt, to get the spoken word example to work, I found that I really need a DMA example for the mic, so that the CPU doesn’t spend all its time sampling the mic. I’m not familiar with SAMD DMA, so it’s not something I can take on right now.
When I asked Seeed about this, they sent me to this tutorial: https://wiki.seeedstudio.com/Wio-Terminal-Audio-Overview/. It looks like they’re using I2S to record data, which is great, but it requires separate hardware. If you know of any ADC DMA examples, that would help a lot!