Nordic AI IoT Axon NPU Speech Recognition MEMS Microphone Demo (NRF54LM20) Dev Kit Part 3

Embedded Development Boards, Kits, Programmers
1

This article will include more details of the previous NPU Speech Recognition MEMS Microphone Demos (Part 1 and Part 2) for the award winning development IoT kit from Nordic NRF54LM20-DK Nordic Semiconductor ASA | Development Boards, Kits, Programmers | DigiKey

image

In order to setup the PDM MEMS microphone eval board, proceed to connect it to the new award winning development AI IoT kit from Nordic NRF54LM20-DK,

image

as follows,

 Adafurit PDM MEMS MIC - nRF54LM20 DK
  • 3V - VDD:IO
  • GND - GND
  • SEL - GND
  • CLK - P1.4
  • DAT - P1.5

In order to build this demo please refer to our previous Zephyr installation article. After completing those steps then open the Python Virtual Environment as usual,

@DigiKey_Coffee_Cup /zephyr$ source venv/bin/activate
   
@DigiKey_Coffee_Cup   (venv) /zephyr$ nrfutil sdk-manager toolchain launch --ncs-version v3.3.0-preview2 --shell

Copy the existing directory edge-ai/applications/ww_kws_new to a new directory called edge-ai/applications/ww_kws_new1.

Please create a new account in the following Nordic Edge AI website.. After creating the account, the steps are:

  1. Create the Wake Word as needed (In this case Change LED was used)
  1. Train the AI model (It can take an hour or so)
  1. Deploy into the firmware of the nRF54LM20 DK

Now use the firmware into the directory nrf_edgeai_generated,


├── nrf_edgeai_generated
│   ├── nrf_edgeai_user_model_axon.h
│   ├── nrf_edgeai_user_model.c
│   ├── nrf_edgeai_user_model.h
│   ├── nrf_edgeai_user_types.h
│   └── prj_example.conf

Modify the wakeword.c file in line 23 as folllows

int ww_init(void)
{
	//ww_model = nrf_edgeai_user_model_wakeword();
	ww_model = nrf_edgeai_user_model_92515();
        ....
        ...

Also modify the Artificial Intelligence Axon Neural Network Sigmoid Activation function to the new one used. (This could be updated later after the completion of this article and these DigiKey Coffee Cup modifications might not be needed.)

//extern nrf_axon_result_e nrf_axon_nn_op_extension_sigmoid_v2(void* , uint16_t , void* );
	
extern nrf_axon_result_e nrf_axon_nn_op_extension_sigmoid(void* , uint16_t , void* );

as needed and then compile it as follows,

(v3.3.0-preview2) @DigiKey_Coffee_Cup/zephyr$ west build -p always -b nrf54lm20dk/nrf54lm20b/cpuapp edge-ai/applications/ww_kws_new1/

...
...
[281/281] Linking C executable zephyr/zephyr.elf
Memory region         Used Size  Region Size  %age Used
           FLASH:      151448 B      1940 KB      7.62%
             RAM:       26906 B       511 KB      5.14%
        IDT_LIST:          0 GB        32 KB      0.00%

After flashing the board with the new firmware, the LED will activate after the Change LED is recognized by the AI IoT Neural Processing Unit as shown in the next video,

Also if a minicom terminal is opened the following information should be displayed after the Change LED is recognized by the Nordic Artificial Intelligence Neural Processing Unit (NPU),

 @DigiKey_Coffee_Cup# minicom -D /dev/ttyACM0
   
   
   Welcome to minicom 2.10

OPTIONS: I18n 
Port /dev/ttyACM0, 08:28:21 [U]

Press CTRL-A Z for help on special keys

Wakeword detected

This has completed the AI IoT Speech Recognition demonstration for this new Nordic AI IoT platform. Please stay tuned for our next part in this series. This new Embedded World 2026 award winning Nordic AI Axon NPU IoT NRF54LM20-DK development kit is available at DigiKey.

Have a great day!

This article is available in spanish here.

Este artículo está disponible en español aquí.

1 Like
Nordic AI IoT Axon NPU Reconocimiento de Frase con Micrófono MEMS (NRF54LM20) Dev Kit Parte 3