CubeAI Studio: E200 ValidationError with Audio Event Detection from model zoo

[dineshsaravanan]

Hardware

• Board: NUCLEO‑U5A5ZJ‑Q
• MCU: STM32U5A5
• ST‑LINK

Software / Tools

• ST Edge AI (via STM32CubeAI Studio)
• Target: STM32U5
• Model source: STM32 AI Model Zoo
• Model used:
  miniresnetv2_s1_64x50_tl_int8.tflite
  (Audio Event Classification → ESC‑10)
  
  

  Steps Followed

  1. Imported the miniresnetv2_s1_64x50_tl_int8.tflite model into STM32CubeAI Studio
  2. Selected NUCLEO‑U5A5ZJ‑Q as the target board
  3. Kept default settings in all steps
  4. Run on Target
     Generated code, built, flashed, and attempted (Step 4 – Validate) failed 

  ---

  Observed Issue

  During Run on Target, validation fails with the following error:

  Run failed: Error: validate error:
  E200 (ValidationError):
  TARGET: Unable to bind the ST.AI runtime with "network" c-model
  E801 (HwIOError): Invalid firmware - COM8:115200

  I have attached the complete validation failure log for reference.

  Additional Observation

  During Step 1 (Code Generation), I consistently receive the following warning from STM32CubeMX:

  Warning: Code Generation
  Main Config: These peripherals still have some not configured or wrong parameter values:
  [Clock]

  I proceeded with code generation despite the warning.

  Any guidance or suggestions to resolve this issue would be greatly appreciated.

  Thank you in advance for your support.

  Best regards,
  Dinesh

[Julian E.]

Hi @dineshsaravanan,

I confirm the issue. I am looking internally for information

Have a good day,

Julian

[Julian E. · Accepted solution]

Hi @dineshsaravanan,

The issue is caused by MX, it seems that the clock config was wrong for some reason.

Here is a fix that should slove it.

Pease go to: C:\ST\STM32Cube AI Studio\STM32Cube-AI-Studio\resources\stedgeai-desktop-backend\dependencies\ 

and open to edit: boardExternalMem.xml

Then look for the die id 481, around line 3800 and replace the current config by this:

        <die id="481">
            <testOptimal>
            RCC.HCLKFreq_Value=160000000
            </testOptimal>
            <optimalParameters>
            RCC.SYSCLKSource=RCC_SYSCLKSOURCE_PLLCLK
            RCC.PLLM=1
            RCC.PLLN=80
            RCC.PLL1R=2
            RCC.PLL2FRACN=0
            RCC.PLL3FRACN=0
            RCC.PLLFRACN=0
            RCC.PLLSourceVirtual=
            RCC.HSE_VALUE=16000000
            RCC.MSIClockRange=RCC_MSIRANGE_4
            RCC.SYSCLKFreq_VALUE=160000000
            RCC.VCOInputFreq_Value=4000000
            RCC.VCOInput2Freq_Value=4000000
            RCC.VCOInput3Freq_Value=4000000
            RCC.USBPHYCLockSelection=RCC_USBPHYCLKSOURCE_HSE
            RCC.Mode=HSE-External-Oscillator
            </optimalParameters>
        </die>

You should have something like this:

This will be fixed in the next release of CubeAI Studio but please edit it manually if you need it before that.

Let me know if this helped.

Have a good day,

Julian

[dineshsaravanan]

Thanks for the support.
After applying the suggested steps, I’m now able to see the Inference Report during AED model validation on target. The issue is resolved.

Appreciate the help!