Rework Voice, Microphone, and MWW handling

Needed for recent changes in ESPHome and to properly align us with the
most recent developments. Includes support for multiple selectable wake
words as well.

supersensor.yaml

@@ -27,6 +27,7 @@ esphome:
   project:
     name: joshuaboniface.supersensor
     version: "2.0"
+  min_version: 2025.5.0
   on_boot:
     - priority: 600
       then:
@@ -38,19 +39,9 @@ esphome:
         - light.turn_on:
             id: output_led
             effect: flash_white
-    - priority: -600
+
-      then:
+preferences:
-        - wait_until:
+  flash_write_interval: 15sec
-            api.connected:
-        - delay: 5s
-        - if:
-            condition:
-              switch.is_on: enable_voice_support
-            then:
-              - logger.log: "Initializing voice assistant on boot"
-              - switch.turn_off: voice_support_active
-              - delay: 2s
-              - switch.turn_on: voice_support_active
 
 dashboard_import:
   package_import_url: github://joshuaboniface/supersensor2/supersensor.yaml
@@ -59,8 +50,6 @@ esp32:
   board: esp32dev
   framework:
     type: esp-idf
-    version: 4.4.8
-    platform_version: 5.4.0
     sdkconfig_options:
       CONFIG_ESP32_DEFAULT_CPU_FREQ_240: "y"
       CONFIG_ESP32_DEFAULT_CPU_FREQ_MHZ: "240"
@@ -259,28 +248,28 @@ script:
             id(supersensor_occupancy).publish_state(new_state);
           }
 
-preferences:
-  flash_write_interval: 15sec
-
 logger:
   level: INFO
   baud_rate: 115200
 
 api:
   reboot_timeout: 15min
-  services:
+  on_client_connected:
-    - service: restart_voice_assistant
+    - script.execute: light_off
-      then:
+    - if:
-        - logger.log: "Manually restarting voice assistant"
+        condition:
-        - voice_assistant.stop:
+          - switch.is_on: enable_voice_support
-        - delay: 2s
+        then:
-        - if:
+          - micro_wake_word.start:
-            condition:
+  on_client_disconnected:
-              switch.is_on: enable_voice_support
+    - light.turn_on:
-            then:
+        id: output_led
-              - switch.turn_off: voice_support_active
+        effect: flash_white
-              - delay: 1s
+    - if:
-              - switch.turn_on: voice_support_active
+        condition:
+          - switch.is_on: enable_voice_support
+        then:
+          - micro_wake_word.stop:
 
 ota:
   platform: esphome
@@ -307,44 +296,7 @@ time:
       then:
         - logger.log: "Time synchronized with Home Assistant"
 
-uart:
-  id: ld2410_uart
-  rx_pin: GPIO19
-  tx_pin: GPIO18
-  baud_rate: 256000
-  data_bits: 8
-  stop_bits: 1
-  parity: NONE
-
-i2c:
-  sda: GPIO27
-  scl: GPIO26
-  scan: true
-
-i2s_audio:
-  i2s_lrclk_pin: GPIO17 # WS
-  i2s_bclk_pin: GPIO16  # SCK
-
-microphone:
-  - platform: i2s_audio
-    id: mic
-    adc_type: external
-    i2s_din_pin: GPIO4 # SD
-    pdm: false
-
 interval:
-  - interval: 5s
-    then:
-      - if:
-          condition:
-            and:
-              - switch.is_on: enable_voice_support
-              - switch.is_on: voice_support_active
-              - not: voice_assistant.is_running
-          then:
-            - logger.log: "Voice assistant not running but should be; restarting"
-            - voice_assistant.start_continuous:
-
   # Regular state reporting to HASS
   - interval: 30s
     then:
@@ -364,59 +316,72 @@ interval:
             App.safe_reboot();
           }
 
+uart:
+  id: ld2410_uart
+  rx_pin: GPIO19
+  tx_pin: GPIO18
+  baud_rate: 256000
+  data_bits: 8
+  stop_bits: 1
+  parity: NONE
+
+i2c:
+  sda: GPIO27
+  scl: GPIO26
+  scan: true
+
+i2s_audio:
+  - id: i2s_input
+    i2s_lrclk_pin:
+      number: GPIO17 # WS
+    i2s_bclk_pin:
+      number: GPIO16  # SCK
+
+microphone:
+  - platform: i2s_audio
+    id: mic
+    i2s_audio_id: i2s_input
+    i2s_din_pin: GPIO4 # SD
+    adc_type: external
+    pdm: false
+    channel: left
+
 micro_wake_word:
+  id: mww
+  microphone:
+    microphone: mic
+    gain_factor: 31
+  stop_after_detection: false
   models:
     - model: hey_jarvis
+      id: mww_hey_jarvis
+    - model: hey_mycroft
+      id: mww_hey_mycroft
+    - model: ok_nabu
+      id: mww_okay_nabu
+    - model: alexa
+      id: mww_alexa
+  vad:
   on_wake_word_detected:
-    then:
+    - logger.log: "A wake word was detected!"
-      - voice_assistant.start:
+    - if:
-          wake_word: !lambda return wake_word;
+        condition:
-
+          voice_assistant.is_running:
-# Include the Espressif Audio Development Framework for VAD support
+        then:
-esp_adf:
+          voice_assistant.stop:
-external_components:
+    - voice_assistant.start:
-  - source: github://pr#5230
+        wake_word: !lambda return wake_word;
-    components:
-      - esp_adf
-    refresh: 0s
 
 voice_assistant:
+  id: va
   microphone: mic
+  micro_wake_word: mww
   use_wake_word: false
-  noise_suppression_level: 3
+  noise_suppression_level: 0
-  auto_gain: 31dBFS
+  auto_gain: 31 dbfs
-  volume_multiplier: 8.0
+  volume_multiplier: 8
-  id: assist
-  on_error:
-    - logger.log: "voice error"
-    - if:
-        condition:
-          and:
-            - switch.is_on: voice_support_active
-            - not: voice_assistant.is_running
-        then:
-          - voice_assistant.start_continuous:
-  on_end:
-    - logger.log: "voice ended"
-    - if:
-        condition:
-          and:
-            - switch.is_on: voice_support_active
-            - not: voice_assistant.is_running
-        then:
-          - voice_assistant.start_continuous:
-  on_client_connected:
-    - light.turn_off:
-        id: output_led
-        transition_length: 2s
-    - script.execute: light_off
-    - lambda: |-
-        id(voice_support_active).publish_state(true);
-  on_client_disconnected:
-    - light.turn_on:
-        id: output_led
-        effect: flash_white
   on_wake_word_detected:
+    - logger.log: "Wake word detected in VA pipeline"
     - light.turn_on:
         id: output_led
         brightness: 100%
@@ -424,6 +389,7 @@ voice_assistant:
         green: 0
         blue: 1
   on_listening:
+    - logger.log: "Listening for commands"
     - light.turn_on:
         id: output_led
         brightness: 100%
@@ -431,14 +397,13 @@ voice_assistant:
         green: 0
         blue: 1
   on_stt_vad_end:
+    - logger.log: "Processing STT result"
     - light.turn_on:
         id: output_led
         brightness: 75%
         red: 0
         green: 1
-        blue: 1
+        blue: 1    
-  on_stt_end:
-    - script.execute: light_off
   on_tts_start:
     - if:
         condition:
@@ -685,16 +650,6 @@ sensor:
     update_interval: 5s
     entity_category: diagnostic
 
-  - platform: template
-    name: "ESP32 CPU Frequency"
-    icon: mdi:cpu-32-bit
-    accuracy_decimals: 1
-    unit_of_measurement: MHz
-    update_interval: 5s
-    lambda: |-
-      return ets_get_cpu_frequency();
-    entity_category: diagnostic
-
   - platform: template
     name: "ESP32 Free Memory"
     icon: mdi:memory
@@ -718,7 +673,9 @@ binary_sensor:
   - platform: gpio
     name: "PIR GPIO"
     id: pir_gpio
-    pin: GPIO32
+    pin:
+      number: GPIO32
+      mode: INPUT_PULLUP
     internal: false
     device_class: motion
     on_press:
@@ -789,26 +746,9 @@ switch:
     optimistic: true
     restore_mode: RESTORE_DEFAULT_OFF
     on_turn_on:
-      - switch.turn_on: voice_support_active
+      - micro_wake_word.start:
     on_turn_off:
-      - switch.turn_off: voice_support_active
+      - micro_wake_word.stop:
-
-  # Active voice support flag/switch
-  - platform: template
-    name: "Voice Support Active"
-    icon: mdi:account-voice
-    id: voice_support_active
-    optimistic: true
-    restore_mode: ALWAYS_OFF
-    entity_category: config
-    on_turn_on:
-      - lambda: id(assist).set_use_wake_word(true);
-      - voice_assistant.stop:
-      - delay: 1s
-      - voice_assistant.start_continuous:
-    on_turn_off:
-      - voice_assistant.stop:
-      - lambda: id(assist).set_use_wake_word(false);
 
   # Global enable/disable for presence LED
   - platform: template
@@ -832,12 +772,12 @@ switch:
 
 number:
   # Temperature offset:
-  # A calibration from -10 to +5 for the temperature sensor
+  # A calibration from -30 to +5 for the temperature sensor
   - platform: template
     name: "Temperature Offset"
     id: temperature_offset_setter
-    min_value: -10
+    min_value: -30
-    max_value: 5
+    max_value: 10
     step: 0.1
     lambda: |-
       return id(temperature_offset);
@@ -848,12 +788,12 @@ number:
             value: !lambda 'return float(x);'
 
   # Humidity offset:
-  # A calibration from -10 to +10 for the humidity sensor
+  # A calibration from -20 to +20 for the humidity sensor
   - platform: template
     name: "Humidity Offset"
     id: humidity_offset_setter
-    min_value: -10
+    min_value: -20
-    max_value: 10
+    max_value: 20
     step: 0.1
     lambda: |-
       return id(humidity_offset);
@@ -951,7 +891,6 @@ number:
         name: "LD2410C Gate8 Still Threshold"
 
 select:
-
   # Occupancy Detect Mode:
   # This selector defines the detection mode for the integrated occupancy sensor. Depending on the
   # selected option, only the given sensor(s) will be used to judge when occupancy begins (i.e.
@@ -1080,6 +1019,39 @@ select:
     distance_resolution:
       name: "LD2410C Distance Resolution"
 
+  - platform: template
+    name: "Wake word sensitivity"
+    optimistic: true
+    initial_option: Moderately sensitive
+    restore_value: true
+    entity_category: config
+    options:
+      - Slightly sensitive
+      - Moderately sensitive
+      - Very sensitive
+    on_value:
+      # Sets specific wake word probabilities computed for each particular model
+      # Note probability cutoffs are set as a quantized uint8 value, each comment has the corresponding floating point cutoff
+      # False Accepts per Hour values are tested against all units and channels from the Dinner Party Corpus.
+      # These cutoffs apply only to the specific models included in the firmware: okay_nabu@20241226.3, hey_jarvis@v2, hey_mycroft@v2
+      lambda: |-
+        if (x == "Slightly sensitive") {
+          id(mww_jarvis).set_probability_cutoff(247);       // 0.97 -> 0.563 FAPH on DipCo (Manifest's default)
+          id(mww_hey_mycroft).set_probability_cutoff(253);  // 0.99 -> 0.567 FAPH on DipCo
+          id(mww_okay_nabu).set_probability_cutoff(217);    // 0.85 -> 0.000 FAPH on DipCo (Manifest's default)
+          id(mww_alexa).set_probability_cutoff(217);        // 0.85 -> 0.000 FAPH on DipCo (Manifest's default)
+        } else if (x == "Moderately sensitive") {
+          id(mww_hey_jarvis).set_probability_cutoff(235);   // 0.92 -> 0.939 FAPH on DipCo
+          id(mww_hey_mycroft).set_probability_cutoff(242);  // 0.95 -> 1.502 FAPH on DipCo (Manifest's default)
+          id(mww_okay_nabu).set_probability_cutoff(176);    // 0.69 -> 0.376 FAPH on DipCo
+          id(mww_alexa).set_probability_cutoff(176);        // 0.69 -> 0.376 FAPH on DipCo
+        } else if (x == "Very sensitive") {
+          id(mww_hey_jarvis).set_probability_cutoff(212);   // 0.83 -> 1.502 FAPH on DipCo
+          id(mww_hey_mycroft).set_probability_cutoff(237);  // 0.93 -> 1.878 FAPH on DipCo
+          id(mww_okay_nabu).set_probability_cutoff(143);    // 0.56 -> 0.751 FAPH on DipCo
+          id(mww_alexa).set_probability_cutoff(143);        // 0.56 -> 0.751 FAPH on DipCo
+        }
+
 text_sensor:
   - platform: wifi_info
     ip_address: