search

tensorflow / tflite-micro

Infrastructure to enable deployment of ML models to low-power resource-constrained embedded targets (including microcontrollers and digital signal processors).
Apache License 2.0
1.75k stars 771 forks source link

Sound (not speech) recognition on esp32-s3-eye v2.2 #2461

Closed gamename closed 4 months ago

gamename commented 4 months ago

Hi,

I am trying to do sound recognition (specifically, a cat meowing). My platform is the esp32-s3-eye. Is this the best library to use? I already have the TinyML model.

Thanks, -T

ddavis-2015 commented 4 months ago

@gamename Hello, and thank you for your interest in TensorFlow Lite for Microcontrollers (TFLM). You have several approaches you can use for your esp32-s3 platform:

  1. Integrate with the existing TFLM repository. After cloning the repo, you could add the appropriate esp32-s3 support in these directories:
    • tensorflow/lite/micro/tools/make/targets
      • your esp32-s3 specific makefile parameters
    • tensorflow/lite/micro/
      • your esp32-s3 specific debug_log.cc and system_setup.cc (examples such as the ceva and chre directories)
  2. Create your own build environment using CMake and the project_generation scripts in:
    • tensorflow/lite/micro/tools/project_generation
      • use the create_tflm_tree script with an Xtensa target architecture (P6?)
      • use the Makefile according to the instructions inside

This approach will probably require quite a bit of work to get to compile correctly.

This issue will now be closed. Please feel free to reopen the issue should you have additional comments.

gamename commented 4 months ago

@ddavis-2015

This issue will now be closed. Please feel free to reopen the issue should you have additional comments.

THank you!

ddavis-2015 commented 4 months ago

@gamename

Suggesting the TFLM Arduino repo. was not correct. The TFLM Arduino repo. is specific to ARM Cortex processors.

However, Espressif has their own TFLM repository which they keep synched with the official Google TFLM repo. The Espressif TFLM code supports the esp32-s3 and can be found here.

gamename commented 4 months ago

@gamename

Suggesting the TFLM Arduino repo. was not correct. The TFLM Arduino repo. is specific to ARM Cortex processors.

However, Espressif has their own TFLM repository which they keep synched with the official Google TFLM repo. The Espressif TFLM code supports the esp32-s3 and can be found here.

Thanks!