Unable to target Nicla Voice board for deployment

Question/Issue:

Unable to target deployment for Arduino Nicla Voice Board (RC Command example)

Project ID:

Context/Use case:

Hello. Uploaded the mcu / dsp / model firmware onto the board using Powershell. Used:
edge-impulse-daemon
to latch the board onto Edge Impulse → now remote management button is green.
The project appears to be built with all green buttons → want to create the binary to build for this target. Stuck that only the Syntiant ML board is showing in the list. The Arduino Nicla Voice is not an available option. What are we doing wrong? Is there a ‘ready to test’ binary for this target board we can upload?

Thanks.

What example of tutorial are you following?

It doesn’t look like the Nicla Voice is fully supported.

Fully supported Arduino boards are:

Thank you for your reply. After fidgeting with the toolchain for the 2nd day…it appears to be now detected.

There are many gotchas in the present documentation. Once I know 100%, will update this thread on how we got it to work. It may have something to do with the execution of the (windows) batch file that must be ran under Powershell to reflash the nicla board with the latest 3 x synpkg files.

At this time, unable to load the ALEXADEMO that is a factory demo using the Arduino toolchain. Led is blinking red, regardless of the upload of the compiled binary. Also, this afternoon, received an update to upgrade the libraries / board support and one of the upgrades was linked to Nicla Voice. Just a FYI.

Got excited too soon. WTH. This thing is going to see the heal of my shoe very soon. No idea what this means…

I may have to rebuild this project and perhaps is missing the extractor section of this project.

The above was a ‘cloned’ example (RC example) → selected this board as the target → then just skipped to the Deployment to build the project. Guessing, in hindsight, that are no shortcuts and the posted example is not really ready to use and needs to be compiled. Will try again…

Hi @mon2 ,

this is the correct project for the Nicla Voice: Syntiant-RC-Go-Stop Nicla - Dashboard - Edge Impulse can you try with this?

regards,
fv

1 Like

Hi @mon2,

Check that the extractor in the Syntiant DSP block is set to log-bin/NDP120, and then regenerate features/retrain the model.
Only the corresponding Syntiant board should be enabled in the deployment block but a recent release seems to have removed the filter.

Aurelien

1 Like

Thanks guys. Getting closer. Can you please assist?

Steps taken:

  1. cloned the Nicla specific project for the RC-GO-Stop demo into my workspace.
  2. Ran the (Matt) Daemon script with --clean command to unlink this Nicla Voice kit away from the older project → then saw a list of potential projects for a new link → selected this new imported project for Nicla Voice → saw the pairing in the project → proceeded to BUILD for this target.
  3. After some time → a .zip file was presented → see attached.
  4. Viewed these files with PowerShell. Attempted to run .\install-lib-win which raised an error of a missing arduino-cli. Downloaded again a fresh copy of this arduino-cli tool → installed it again → performed some upgrades of mbed, etc. (used update and upgrade commands).

Ran again the above .\install-lib-win which raised the same missing arduino-cli tool → just copied and pasted inside the same folder → no more errors.

The tool ran and appears to have downgraded some of the code versions but filed it as a no care.

Tool ran to completion.

Next ran .\flash_windows which performs some updates / downgrades → raised the serial error that is attached.

Launced Teraterm → connected @ 115200 with COM29 (my box) → see the attached output that the model package is missing.

Attempted again to launch via .\flash_windows_model → raises the missing serial message.

What are we doing wrong?

and the attached.

In searching google:

python - No module named serial - Stack Overflow

Results in the same as per attachment. Stuck.

Miracles never cease. Just got it working. Have a better chance of a date with Dua Lipa than to get this working the first time.

Resolution:

  1. Ripped out (uninstall) everything to do with python in my Windows box.
  2. Downloaded the latest version of python in my Windows box.
  3. Shelled in PowerShell in ADMIN mode.
  4. Confirmed python is installed with:

python ; all good

  1. Installed the serial support which was the last hurdle of this painful process:

pip install pyserial

  1. moved to the folder (with Powershell) where we extracted the target fileset for the Nicla Voice board that allows for the GO / STOP voice demo.

  2. Launched the following (in order):

install_lib_win
flash_windows

Resulted in the following output - all good and now working !!!

Will go purchase a lottery ticket and phone up Dua Lipa.

1 Like

and the following output after flashing the board:

1 Like

let us know about the lottery :wink:

1 Like