Does anyone have micro_wake_word working with media_player on esp32 wrover?

jfthomps · September 25, 2024, 12:57am

Does anyone have micro_wake_word working on an esp32 wrover board with media_player instead of speaker? I’ve tried various combinations of things and been able to get micro_wake_word working with a speaker component instead of media_player, and been able to get media_player working without micro_wake_word, but I’ve not been able to get them working together. I’ve see the esphome_audio project at GitHub - gnumpi/esphome_audio: Custom audio components for ESPHome, and it looks like it should do what I want, but I haven’t been able to get the wake word part working.

Arh · September 25, 2024, 6:24am

You really need an esp32-s3-N16R8 board to get this working, if its not an s3 board there will be very little chance of getting it to run. I am no expert though.

The page you link to will not work successfully either, you need to use the code from BigBobbas. Recently updated to cure the restarting issues.

github.com

BigBobbas/esphome_firmware/blob/6e7057d1750609fb0c14c1dee948fed0722f70fa/ESPHome Diy Voice Builds/esp32-s3-adf-mww.yaml

esphome:
  name: esp-speaker
  platformio_options:
    board_build.flash_mode: dio
  on_boot:
    - light.turn_on:
        id: led_ww
        blue: 100%
        brightness: 60%
        effect: fast pulse
esp32:
  board: esp32-s3-devkitc-1
  variant: esp32s3
  framework:
    type: esp-idf
    version: recommended

    sdkconfig_options:
        CONFIG_ESP32_S3_BOX_BOARD: "y"
        CONFIG_ESP32_WIFI_STATIC_RX_BUFFER_NUM: "16"

This file has been truncated. show original

I have it working successfully on 3 setups right now.

jfthomps · September 26, 2024, 11:29pm

Thanks for that information.

Do you know why an s3 board is needed? I know I had micro_wake_word working (without media_player) at some point on my wrover board, though I’m having trouble getting it working again. Does the adf_pipeline module not work with wrover boards?

Arh · September 27, 2024, 6:20am

Sorry no idea on the technicalities, I am sure someone else will know. But chances are its down to available memory.

I have had Voice assistant working with speaker on the wroover in the past but not used that for a long time now so can’t comment on the present state.