Error "Arena size is too small for all buffers" while inferencing audio project on XIAO ESP32-S3

Help
#1

Hi, I was trying to inference an audio KWS project on XIAO ESP-S3 with Sense HAT. The device was working fine on the previous audio project, only facing issues with this particular project. Thanks in Advance.

ERR: Failed to run classifier (-3)
Arena size is too small for all buffers. Needed 15280 but only 12160 was available.
AllocateTensors() failed
ERR: Failed to run classifier (-3)
Edge Impulse Inferencing Demo
Inferencing settings:
Interval: 0.062500 ms.
Frame size: 16000
Sample length: 1000 ms.
No. of classes: 2

a

#2
  • Your MCU is out of memory.
    • Run EON and select a Model with lower memory requirements.
    • Run Performance Calibration to analyze performance scenarios.
#3

Thanks @MMarcial for the response.

Unfortunately, XIAO ESP32-S3 is not supported by EON. So can’t use that right now.

I will check this, thanks.

#4

@shawn_edgeimpulse

  • To reduce the size of an image classification Impulse one can reduce the Image Width and Image Height in the Create Impulse page of the Input Block.

  • What options do we have to reduce the size of an audio Impulse?

#5

Hi @MMarcial,

For audio projects, you can try:

  • Reducing the bit depth (e.g. from 16-bit to 8-bit)
  • Reducing the sampling rate (8 kHz is probably the lowest you want to go for voice data)
  • Using int8 quantized models instead of float 32
  • Using the EON Compiler (if available)
1 Like
#6

@shawn_edgeimpulse Thanks, but unfortunately I was not able to solve it after reducing the bit depth and reducing the sample rate. And the EON is not supported by the XIAO ESP32-S3 now.

#7

As an alternate solution, you can find a MCU that will run the model with the EI Profiler.

#8

@MMarcial , Can you please check the link? Thanks.

#9

@salmanfarisvp

The link takes you to a Hugging Face tutorial. Do not worry about that. Scroll down to the Profile Your Model section.

You can also get to it here:

image
image660×704 71.8 KB

#10

Hi @salmanfarisvp,

Also, what is your window size? If that’s too big, you will run out of memory quickly. I generally find that a window size of 1 sec is the most you will be able to use for MCU KWS projects.

As @MMarcial mentioned, you can use the profiler to get an estimate of the RAM usage. You can use the Python SDK, but it’s not necessary. You can also view RAM/ROM estimations in Studio. Go to the Deployment page. Under Model Optimizations, click Change target and select the Espressif ESP-EYE (the closest board we have to the ESP32-S3).

Screenshot 2023-09-05 at 9.22.20 AM
Screenshot 2023-09-05 at 9.22.20 AM794×256 28.3 KB

Try toggling EON Compiler and viewing the difference between int8 and float32 versions. You should see an estimated total RAM and ROM usage.

Screenshot 2023-09-05 at 9.21.53 AM
Screenshot 2023-09-05 at 9.21.53 AM682×608 75.8 KB

What’s weird is that your ESP32 is reporting that only ~12 kB is available for the arena size. That doesn’t seem right, considering the ESP32-S3 has something like 512kB of RAM. Paging @AIWintermuteAI – any thoughts on this one?

#11

From @AIWintermuteAI: ESP32-S3 optimized kernel support is in the works. Something you might want to try: swap the Edge Impulse C++ kernel with an ESP-NN kernel to support faster/efficient computations on the ESP32-S3. This is a bit of a hack, but it might be worth a shot: TinyML Made Easy: Image Classification - Hackster.io. Note that it will not work with a library created with the EON Compiler enabled.

#12

Yes, mine is 1 sec.

image
image878×632 86.1 KB

it’s 41.6K RAM usage in classification.

Thanks and Yes, I tried to use the ESP-NN lib compiled by @AIWintermuteAI. I’ve been using a small Yes/No project with XIAO-ESP32-S3 with the NN lib swapped and it’s working fine, but here I’m trying to build a running faucet demo with Edge Impulse provided running faucet dataset.

Here is my public project link: running_faucet - Dashboard - Edge Impulse

Thanks for the help and comments.

#13

I tried to split the data into 1 sec to experiment and now I get the different error "AllocateTensors() failedERR: Failed to run classifier (-3)" .

Any suggestions. Thanks.

#14

“ESP32-S3 optimized kernel support is in the works.”
Great to hear - we are just working on a sound recognition solution on ESP32-S3 - can u give a rough idea of schedule?
Also we could do a beta testing if helpful.