esphome sendspin ESP32 (somewhat working)

esph-hifi-1.yaml

substitutions:
  version: 1.0.0
  system_name: hifi-1
  friendly_name: HiFi 1
  # Stolen from the ESP32-S3 configuration, probably should be set to "false"?
  task_stack_in_psram: "true"

packages:
  <<: !include_dir_named ./components/common

esphome:
  <<: !include ./boards/.esphome.yaml
  min_version: 2025.12.2
  name_add_mac_suffix: false
  platformio_options:
    board_build.flash_mode: dio

esp32:
  board: mhetesp32minikit
  flash_size: 8MB
  framework:
    type: esp-idf
    version: recommended
    sdkconfig_options:
      CONFIG_BT_ALLOCATION_FROM_SPIRAM_FIRST: "y"
      CONFIG_BT_BLE_DYNAMIC_ENV_MEMORY: "y"

      CONFIG_MBEDTLS_EXTERNAL_MEM_ALLOC: "y"
      CONFIG_MBEDTLS_SSL_PROTO_TLS1_3: "y"

      # Stolen from the ESP32-S3 configuration, probably should not be here?
      CONFIG_SPIRAM_RODATA: "y"
      CONFIG_SPIRAM_FETCH_INSTRUCTIONS: "y"

# Disabled due to a suspicion that it uses RAM we need on the ESP32, works on the -S3 though!
# network:
#   enable_ipv6: true

http_request:

globals:
  - id: current_volume
    type: float
    initial_value: "0.2"
  - id: announcement_triggered
    type: bool
    initial_value: "false"

psram:
  speed: 80MHz
  ignore_not_found: false

external_components:
  - source: github://pr#12253
    components: [mixer]
    refresh: 0s

  - source: github://pr#12254
    components: [resampler]
    refresh: 0s

  - source: github://pr#12256
    components: [audio]
    refresh: 0s

  - source: github://pr#12258
    components: [media_player]
    refresh: 0s

  - source: github://pr#12284
    components: [mdns, sendspin]
    refresh: 0s

  - source: github://pr#12429
    components: [file, http_request, media_source, speaker_source]
    refresh: 0s

# Disabling these (seemingly) improved stability at least a bit? Still crashing though ...
# text_sensor:
#   - platform: sendspin
#     type: title
#     name: Track Title

#   - platform: sendspin
#     type: artist
#     name: Artist

#   - platform: sendspin
#     type: album
#     name: Album

i2s_audio:
  i2s_lrclk_pin: GPIO25
  i2s_bclk_pin: GPIO26

speaker:
  - platform: i2s_audio
    id: i2s_speaker_id
    sample_rate: 48000
    i2s_dout_pin: GPIO22
    bits_per_sample: 16bit
    dac_type: external
    channel: stereo
    timeout: never
    buffer_duration: 100ms

  - platform: mixer
    id: mixer_speaker_id
    output_speaker: i2s_speaker_id
    num_channels: 2
    task_stack_in_psram: ${task_stack_in_psram}
    source_speakers:
      - id: announcement_mixer_input
        timeout: never
      - id: media_mixer_input
        timeout: never

  - platform: resampler
    id: media_resampling_speaker
    output_speaker: media_mixer_input
    sample_rate: 48000
    bits_per_sample: 16

  - platform: resampler
    id: announcement_resampling_speaker
    output_speaker: announcement_mixer_input
    sample_rate: 48000
    bits_per_sample: 16

sendspin:
  id: sendspin_hub
  task_stack_in_psram: ${task_stack_in_psram}
  kalman_process_error: 0.01

media_source:
  - platform: sendspin
    id: sendspin_source
  # Disabled for testing.
  # - platform: http_request
  #   id: http_source
  #   buffer_size: 200000

media_player:
  - platform: sendspin
    id: sendspin_group_media_player

  - platform: speaker_source
    id: external_media_player
    name: "${friendly_name} Player"
    announcement_speaker: announcement_resampling_speaker
    media_speaker: media_resampling_speaker
    announcement_pipeline:
      format: FLAC # FLAC is the least processor intensive codec
      num_channels: 1 # Stereo audio is unnecessary for announcements
      sample_rate: 48000
    media_pipeline:
      format: FLAC # FLAC is the least processor intensive codec
      num_channels: 2
      sample_rate: 48000
    volume_increment: 0.05
    sources:
      # - http_source
      - sendspin_source

    on_announcement:
      - if:
          condition:
            - lambda: return (!id(announcement_triggered));
          then:
            - lambda: |-
                id(announcement_triggered) = true;
                id(current_volume) = id(external_media_player).volume;
            - mixer_speaker.apply_ducking:
                id: media_mixer_input
                decibel_reduction: 40
                duration: 0.0s
            - wait_until:
                media_player.is_announcing: external_media_player
            - delay: 0.75s
            - media_player.volume_set:
                id: external_media_player
                volume: !lambda "return id(announce_volume).state / 100.0;"
            - wait_until:
                - not:
                    media_player.is_announcing: external_media_player
            - mixer_speaker.apply_ducking:
                id: media_mixer_input
                decibel_reduction: 0
                duration: 1.0s
            - media_player.volume_set:
                id: external_media_player
                volume: !lambda "return id(current_volume);"
            - lambda: id(announcement_triggered) = false;

    on_state:
      if:
        condition:
          and:
            - lambda: return (!id(announcement_triggered));
            - not:
                media_player.is_announcing: external_media_player
        then:
          - mixer_speaker.apply_ducking:
              id: media_mixer_input
              decibel_reduction: 0
              duration: 1.0s

number:
  - platform: template
    name: Announce Volume
    id: announce_volume
    icon: mdi:volume-source
    unit_of_measurement: "%"
    min_value: 0
    max_value: 100
    step: 1
    restore_value: true
    initial_value: 50
    optimistic: true

button:
  - platform: restart
    id: restart_btn
    name: "${friendly_name} REBOOT"