Esphome on Atom

Configuration Voice Assistant
1

I have a basic config that is sending and receiving pipeline data. It picked up my tv in background perfectly in continuous mode. I turned off continuous mode for button only testing, but when I try to use the button to activate the mic, and a give it command, HA says "you" and it didn't understand me. here is my yaml. Please suggest what is going on as gemini and grok have me going in circles. btw, the libraries on github don't have the right gpio's for this model. these are accurate.

esphome:
name: "atom1"
friendly_name: atoms3r1

on_boot:
priority: -100
then:
- wait_until: api.connected
- delay: 8s
# - voice_assistant.start_continuous:
- media_player.volume_set:
id: va_media_player
volume: 0.7

esp32:
board: esp32-s3-devkitc-1
variant: esp32s3
flash_size: 8MB
framework:
type: esp-idf
sdkconfig_options:
CONFIG_FREERTOS_HZ: "1000"

psram:
mode: octal
speed: 80MHz

logger:
level: DEBUG

api:
encryption:
key: xxxxxxxxxxxxxxxxxxxxx

ota:

* platform: esphome
  password: xxxxxxxxxxxxxxxxxxxxxxxxxxxxx

wifi:
ssid: !secret wifi_ssid
password: !secret wifi_password
ap:
ssid: "Atom1 Fallback Hotspot"
password: "xxxxxxxxxxxxxxxx"

captive_portal:

# ==================== I2C ====================

i2c:

* id: bus_a
  sda:
  number: GPIO45
  ignore_strapping_warning: true

  scl:
  number: GPIO0
  ignore_strapping_warning: true
  scan: true
  frequency: 50kHz
  timeout: 100ms

# ==================== NS4150B AMPLIFIER ====================

switch:

* platform: gpio
  id: speaker_amp
  name: "Speaker Amplifier"
  pin:
  number: GPIO18 # uses the NS4150 CTR audio amp
  inverted: false
  restore_mode: ALWAYS_ON

# ==================== AUDIO ====================

i2s_audio:

* id: i2s_bus
  i2s_lrclk_pin:
  number: GPIO3 # WS
  ignore_pin_validation_error: true
  i2s_bclk_pin: GPIO17 # SCK
  i2s_mclk_pin: GPIO11 # MCLK

audio_dac:

* platform: es8311
  id: es8311_dac
  i2c_id: bus_a
  address: 0x18
  use_microphone: true
  mic_gain: 0DB          # Lowered for testing

# ==================== SPEAKER ====================

speaker:

* platform: i2s_audio
  id: echo_speaker
  i2s_audio_id: i2s_bus
  i2s_dout_pin: GPIO48
  channel: mono
  dac_type: external
  audio_dac: es8311_dac
  buffer_duration: 200ms      # ← Add this

# ==================== MICROPHONE (Fixed) ====================

microphone:

* platform: i2s_audio
  id: echo_mic
  i2s_audio_id: i2s_bus
  i2s_din_pin: GPIO4
  adc_type: external
  pdm: false
  channel: left          # ← Must be stereo
  sample_rate: 16000
  bits_per_sample: 16bit

# ==================== MEDIA PLAYER ====================

media_player:

* platform: speaker
  id: va_media_player
  name: "Atom1 Feedback"
  announcement_pipeline:
  speaker: echo_speaker
  num_channels: 1
  sample_rate: 16000
  format: FLAC                 # Helps with timing

# ==================== BUTTON ====================

binary_sensor:

* platform: gpio
  pin:
  number: GPIO41
  inverted: true
  mode: input_pullup
  name: "AtomS3R Button"
  filters:
  * delayed_on_off: 50ms
    on_press:
  * switch.turn_on: speaker_amp
  * delay: 150ms
  * voice_assistant.start:

  # Optional: Add an "on_release" to stop if you prefer "Push and Hold"

  #- voice_assistant.stop:

# ==================== VOICE ASSISTANT ====================

voice_assistant:
id: va
microphone: echo_mic
media_player: va_media_player
use_wake_word: false #false use button for now   # <-- true Forces device to wait for your wake word
noise_suppression_level: 2
auto_gain: 15dBFS
volume_multiplier: 2.0

# Plays a chime the moment you press the button

on_listening:
- switch.turn_on: speaker_amp
- delay: 120ms

on_wake_word_detected:
- media_player.play_media:
id: va_media_player
media_url: "https://github.com/esphome/home-assistant-voice-pe/raw/dev/sounds/wake_word_triggered.flac"

# REMOVE THESE:

# on_end:

# - voice_assistant.start_continuous:

# on_error:

# - voice_assistant.start_continuous:
2

What's with all the *'s?

3

what are you asking? the #'s are for rem'ing out the notes or line.

4

Not #... *... almost all of your lists are using * instead of -

image

Also, almost everything is improperly indented.

5

thats odd, it must be the rendering of the page. they are all "-"

6

OK, some progress. the HA is hearing and executing on the commands using the button press. next step is to get the wake work going. here is a working example. The speaker did have static output at times, but I didnt know if HA executes the request, it dosent say anything.

esphome:
  name: atom2
  friendly_name: atom2
  min_version: 2025.5.0
  on_boot:
    priority: 600
    then:
      - speaker.volume_set:
          id: va_speaker
          volume: 80%  # Set to a comfortable level

esp32:
  board: esp32-s3-devkitc-1   # Change if your board has a different variant
  variant: esp32s3
  flash_size: 8MB
  framework:
    type: esp-idf

psram:
  mode: octal
  speed: 80MHz
# ====================== WiFi & Connectivity ======================
wifi:
  ssid: !secret wifi_ssid
  password: !secret wifi_password
  power_save_mode: none
  manual_ip:
    static_ip: 192.168.15.xx   # Use the actual IP you want for atom2
    gateway: 192.168.15.1      # Your router's IP
    subnet: 255.255.255.0
  use_address: 192.168.15.xx   # This tells the dashboard EXACTLY where to look
  ap:
    ssid: "Atom1 Fallback Hotspot"
    password: "Vf9FJJ0KRrkF"


api:
  encryption:
    key: "xxxxx"

ota:
  - platform: esphome
    password: "xxxxx"

logger:
  level: INFO

# ====================== I2C for ES8311 ======================
i2c:
  - id: bus_a
    sda: 
      number: GPIO45  #SDA
      ignore_strapping_warning: true
    
    scl: 
      number: GPIO0 #SCL
      ignore_strapping_warning: true
    scan: true
    frequency: 400kHz
    timeout: 100ms

# ====================== I2S Audio Bus ======================
i2s_audio:
  - id: i2s_bus
    i2s_lrclk_pin: GPIO3     # WS  lrclk
    i2s_bclk_pin: GPIO17    # BCLK
    i2s_mclk_pin: GPIO11      # MCLK

# ====================== ES8311 Audio Codec ======================
audio_dac:
  - platform: es8311
    id: es8311_dac
    i2c_id: bus_a
    address: 0x18               # Most common address for ES8311
    bits_per_sample: 16bit
    sample_rate: 48000

# ====================== Microphone ======================
microphone:
  - platform: i2s_audio
    id: va_mic
    i2s_audio_id: i2s_bus
    i2s_din_pin: GPIO4        # DIN (mic data into ESP)
    adc_type: external
    pdm: false
    channel: left
    bits_per_sample: 32bit     # ES8311 often outputs 32-bit for mic
    sample_rate: 16000

# ====================== Speaker (via ES8311 + NS4150B) ======================
speaker:
  - platform: i2s_audio
    id: va_speaker
    i2s_audio_id: i2s_bus
    i2s_dout_pin: GPIO48        # DOUT 
    dac_type: external
    audio_dac: es8311_dac
    buffer_duration: 1200ms  # Try increasing from default (500ms) to 800ms or 1000ms
    channel: left
    bits_per_sample: 16bit
    sample_rate: 48000

# ====================== Amplifier Control (NS4150B) ======================
switch:
  - platform: gpio
    id: speaker_amp
    name: "Speaker Amplifier"
    pin:
      number: GPIO18          # NS4150_CTR
      inverted: true         # Change to true if speaker doesn't turn on
    restore_mode: ALWAYS_OFF

# ====================== Button (optional) ======================
binary_sensor:
  - platform: gpio
    pin:
      number: GPIO41  # USER_BUT
      mode: INPUT_PULLUP
      inverted: true
    name: "Voice Button"
    filters:
      - delayed_on_off: 50ms
    on_press:
      - voice_assistant.start:

# ====================== Voice Assistant ======================
voice_assistant:
  id: va
  microphone: va_mic
  speaker: va_speaker        # Using speaker component (better latency)

  # Audio tuning
  noise_suppression_level: 2
  auto_gain: 31dBFS
  volume_multiplier: 1.4

  # Amp / behavior control
  on_wake_word_detected:
    - switch.turn_on: speaker_amp

  on_listening:
    - switch.turn_on: speaker_amp
    - delay: 500ms

  on_tts_start:
    - switch.turn_on: speaker_amp

  on_end:
    - delay: 1200ms
    - switch.turn_off: speaker_amp

  on_idle:
    - switch.turn_off: speaker_amp