Good morning to all, my name is Pierluigi and I am writing from Italy.
I would kindly need your support.
I have great experience with the use of ATmega type processors as well as computer science, I am now following a project where I need to create an ‘intelligent VOX’.
In practice I would need a way to understand when someone is speaking, a sort of VOX but more advanced capable of not detecting unwanted noises.
The Edge Impulse ML seems does not support Nicla Voice even in some online tutorial it seemed like it was possible, I bought it for this reason.
Really my goal is prorotype a device able to recognize the human voice, simply if a person is speaking or not, not a single word like ‘Alexa’ or other…
Many online project use the ‘spotting’ triggering, like Alexa demo but I wolud recognize ‘simply’ the human voice 
Is possible to to it eventually?
Thanks a lot for your help.
bst
pierluigi
Hi @pierbert
It is a supported devkit: Arduino Nicla Voice | Edge Impulse Documentation
Can you clarify a bit your use case, and the problems encountered?
I’m not quite getting it @ei_francesco do you know what this use case is?
Best
Eoin
Dear Eoin thanks firs of all for your answer, I will try to explain my problem as best as possible.
My goal is to obtain a system able to recognize the presence of human voice, I don’t need spotting or other feautres, I need only trigger if a people is talking or not, lets’ say similar a ‘smart VOX’.
In order to obtain this (I really either dont’ know if is possibile with this board) I buyed a Nicla Voice.
First of all I tried to deploy a test application found in google, in order to better understand the (not banal) steps in machine learning.
At the end of the all process in the deployment section of edgeimpulse panels Nicla Voice is not present as possible choice, there are only Nicla Vision and another pair that now do not remember exactly. I tried to build the code selecting Nicla Vision (do not having other possibilities) but then during uploading in the board with CLI it returns an error like if the board is not present.
The final question(s) is(are)
-Can I use Nicla Vision as target device even I have a Nicla Voice or are differents?
-Is possible to create this kind of ‘smart VOX’ with Nicla Voice or risk I to waste time witouth obtain my goal.
Thanks a lot again for your patiente and support.
bst
pierluigi
Hi @pierbert
Can I use Nicla Vision as target device even I have a Nicla Voice or are differents?
No, they are totally different target. The Nicla voice has Syntiant NDP 120 to run NN, the Nicla Vision runs NN on a Cortex-M7.
Is possible to create this kind of ‘smart VOX’ with Nicla Voice or risk I to waste time witouth obtain my goal.
You can try following this guide Recognize sounds from audio | Edge Impulse Documentation , but remember to use the Syntiant Audio block, see here Responding to your voice - Syntiant - RC Commands | Edge Impulse Documentation
At the end of the all process in the deployment section of edgeimpulse panels Nicla Voice is not present as possible choice, there are only Nicla Vision and another pair that now do not remember exactly.
This because to use a target with the Syntiant NDP, when setting up your impulse in Studio, you should add one of the 2 Syntiant blocks (for IMU or audio, depends on your project).
regards,
fv
Dear Francesco, thanks first of all for your answer, I will try to follow your tips.
Thanks a lot in advance,
bst,
pierluigi