No output issue from MP23ABS1 on the old version of STWINKT1.

[TKIM.3]

I have two different HW versions of STWIN and there is an analog mic difference; MP23ABS1 vs MP34ABSU. When I test with MP23ABS1, it looks it has some issue that it doesn't output sound data at all. but, the new revision with MP34ABSU hasn't that issue with the same firmware.

My questions are

1. why STWIN has been updated with a new MIC ; MP34ABSU?
2. Does MP23ABS1 some issue? if it has, let me know what kinds of issues it has.

[Eleon BORLINI]

Hi @TKIM.3​ ,

please note that the MP23ABS1 and the IMP23ABSU microphones are almost the same from a firmware point of view.

I'm not sure to have guessed your concern... are you saying that the STWINKT1+MP23ABS1 was not working, while STWINKT1+MP23ABS1 is?

-Eleon

[DKosy.1]

Hello

I have the Sensortile.Box which contains an analog mems microphone (MP23ABS1) and i want to deploy an audio recognition application .In order to interact with the microphone do i need to do ADC ? Have you ever tried audio sampling in Sensortile.box and can give me some pointers ?

[Eleon BORLINI]

Hi @DKosy.1​ ,

there are some examples in the FP-AI-SENSING1 library, that implements the audio acquisition and processing for the onboard MP23ABS1 microphone.

For example, you could start from the application example in \Projects\STM32L4R9ZI-SensorTile.box\Applications\SENSING1.

You can also have a look to the SensorTile.box_audio.c file in the Drivers folder \Drivers\BSP\SensorTile.box for the specific audio data acquisition and processing functions.

The library for the analog microphone data acquisition uses the DFSDM peripheral (Digital Filter for Sigma-Delta Modulator) as a generic ADC.

-Eleon

[DKosy.1]

Hello

I am i trying to build the example application

/STM32CubeFunctionPack_SENSING1_V4.0.2/Projects/STM32L4R9ZI-SensorTile.box/Applications/SENSING1

in stm32cubeide v1.5.1 but i am getting the following error

cannot find -l:NetworkRuntime410_CM4_GCC.a  

Do you know what is causing it ?

I have already the stm32cube-ai library installed .

[Eleon BORLINI]

Hi @DKosy.1​ ,

did you made some progresses this week?

Did you make some modifications in the code before compiling?

Maybe it's better to post your question also on STM32 community page, with STM32-AI topic, for more help.

-Eleon

[DKosy.1]

Hi ,maybe the problem was that i was using stm32cubeide on linux ,when i switched to windows it compiled successfully .

I was looking at the code from the demo that you had ,can i create an audio classification of my own according to the example and use the cube-ai for inference of the model ?

[Eleon BORLINI]

Hi @DKosy.1​ ,

first, glad to hear you solved your previous issue.

You could create an audio classification using -for example- Tensorflow, or similar package: you can then convert it into C code using the X-CUBE-AI tool (here for some more details on STM32Cube.AI ecosystem).

You could base your neural network on audio feature or on spectrogram pattern recognition.

-Eleon