Il Display Definitivo per Home Assistant: Waveshare ESP32-S3 Touch AMOLED
Home Assistant
Scritto da Vincenzo Caputo
Ammettiamolo: tutti noi, almeno una volta, abbiamo guardato Iron Man e pensato: "Voglio quella roba lì in casa mia". Non Alexa, non Google Nest. Voglio Jarvis. Voglio un'interfaccia olografica che pulsa mentre mi risponde e una voce che capisce cosa dico anche se sto ascoltando gli AC/DC al massimo volume.
Così, armati di speranza e saldatore, siamo partiti con il Waveshare ESP32-S3 Touch 1.75inch AMOLED. Bello, bellissimo. Un display IPS, touch capacitivo, processore potente. "Ha anche i microfoni integrati!", abbiamo esclamato.
Spoiler: È stato un disastro!
I microfoni integrati nel Waveshare sono lì più per figura che per funzionalità. Senza un DSP (Digital Signal Processor) dedicato e senza cancellazione dell'eco (AEC), tentare di parlarci mentre rispondeva era come urlare in una galleria del vento. Inoltre, gestire l'elaborazione grafica a 30fps e la pipeline vocale sullo stesso chip portava l'ESP32 a chiedere pietà (e a crashare spesso).
La soluzione? Divide et impera. Abbiamo deciso di operare una chirurgia hardware: separare il "Cervello" (Audio/Logica) dalla "Faccia" (Grafica).
In questa guida vedremo come creare un sistema distribuito usando:
Il Cervello: ReSpeaker Lite (per un riconoscimento vocale impeccabile).
La Faccia: Waveshare ESP32-S3 1.75" (dedicato puramente all'animazione sci-fi).
Il Ponte: Una comunicazione HTTP diretta tra i due dispositivi.
I due dispositivi utilizzati li potete acquistare su Amazon ai seguenti link:
Prezzo intero: 0,00€
Prezzo scontato: 45,99€
Prezzo intero: 0,00€
Prezzo scontato: 58,83€
Oppure sui siti dei rispettivi produttori:
https://www.waveshare.com/product/arduino/displays/amoled/esp32-s3-touch-amoled-1.75.htm
I Prerequisiti
Prima di iniziare a compilare codice come se non ci fosse un domani, assicurati di avere:
Una configurazione funzionante di Home Assistant con Assist (Pipeline configurata con Whisper e Piper, o meglio ancora con LLM integrato).
ESPHome installato e aggiornato.
Aver letto il nostro precedente articolo sul ReSpeaker Lite, perché daremo per scontato che abbiate già flashato la base.
L'Hardware: Perché questa accoppiata?
1. Il ReSpeaker Lite (Le Orecchie)
Questo non è un semplice microfono. Monta un chip XMOS XU-316. In termini tecnici, significa che ha un array di microfoni con cancellazione dell'eco hardware e riduzione del rumore on-board. Può "sentire" la wake-word anche mentre l'altoparlante sta suonando musica a tutto volume (Barge-in support). L'ESP32 integrato qui si occupa solo di gestire l'audio e la connessione con Home Assistant.
2. Il Waveshare ESP32-S3 1.75" (Gli Occhi)
Questo bestione deve gestire un pannello RGB 466×466. L'interfaccia RGB richiede molta RAM e cicli di clock. Liberandolo dal compito di ascoltare ed elaborare l'audio, possiamo usare tutta la potenza del dual-core Xtensa LX7 per disegnare vettori, animazioni fluide e interfacce stile "Arc Reactor" senza lag.
Passo 1: La "Faccia" (Configurazione Waveshare)
L'idea è semplice: il Waveshare non deve "sapere" nulla di Home Assistant o di voice pipeline. Lui deve essere uno "schiavo" che riceve un numero da 0 a 4 e cambia faccia di conseguenza.
Usiamo un'entità number esposta via web_server. Questo ci permette di comandarlo con una semplice richiesta HTTP POST.
Il Codice YAML (Waveshare):
Per la grafica, abbiamo optato per un design "Cyberpunk Geometrico" ottimizzato per non uccidere la CPU (niente immagini pesanti, solo pura matematica vettoriale).
Ricordatevi di aggiornare i campi SSID e Password della vostra rete WiFi
substitutions:
name: "jarvis-face"
friendly_name: "Jarvis Face Display"
esphome:
name: "${name}"
friendly_name: "${friendly_name}"
# Opzioni specifiche per la board Waveshare S3
platformio_options:
board_build.flash_mode: dio
board_build.f_flash: 80000000L
board_build.f_cpu: 240000000L
esp32:
variant: esp32s3
flash_size: 16MB
framework:
type: esp-idf
psram:
mode: octal
speed: 80MHz
wifi:
ssid: "tuossid" #qui metti all'interno delle virgolette la tua rete WiFi
password: "tuapasswordwifi" #qui metti all'interno delle virgolette la tua password WiFi
api:
reboot_timeout: 0s
ota:
platform: esphome
logger:
globals:
- id: jarvis_state
type: int
initial_value: '0'
restore_value: false
# =========================
# WEB SERVER (Ricevitore Comandi)
# =========================
web_server:
port: 80
number:
- platform: template
id: jarvis_state_control
name: "jstate" # <--- Nome corto e semplice
optimistic: true
min_value: 0
max_value: 4
step: 1
on_value:
- lambda: |
id(jarvis_state) = (int)x;
ESP_LOGI("HTTP", "Stato ricevuto: %d", (int)x);
# =========================
# BUS CONFIGURATION (QSPI)
# =========================
spi:
- id: display_qspi
type: quad
clk_pin: GPIO38
data_pins:
- GPIO4
- GPIO5
- GPIO6
- GPIO7
# =========================
# DISPLAY (Piattaforma MIPI SPI)
# =========================
display:
- platform: mipi_spi
id: disp1
model: CO5300
bus_mode: quad
reset_pin: GPIO39
cs_pin: GPIO12
dimensions:
height: 466
width: 466
update_interval: 33ms
lambda: |-
it.fill(Color(0,0,0));
int s = id(jarvis_state);
long t = millis();
int cx = 233; int cy = 233;
// --- COLORI BASE ---
Color ciano = Color(0, 255, 240);
Color viola = Color(160, 30, 255);
Color rosso = Color(255, 50, 50);
Color bianco = Color(255, 255, 255);
Color verde = Color(0, 255, 100);
Color main_color = ciano;
if (s == 2) main_color = viola;
if (s == 3) main_color = verde;
if (s == 4) main_color = rosso;
// --- 1. HUD CORNICE (Nuovo!) ---
// Disegna 4 angolari agli estremi per dare struttura "Visore"
int d = 10; // Distanza dal bordo
int l = 40; // Lunghezza linea
// Alto Sx
it.line(d, d, d+l, d, ciano); it.line(d, d, d, d+l, ciano);
// Alto Dx
it.line(466-d, d, 466-d-l, d, ciano); it.line(466-d, d, 466-d, d+l, ciano);
// Basso Sx
it.line(d, 466-d, d+l, 466-d, ciano); it.line(d, 466-d, d, 466-d-l, ciano);
// Basso Dx
it.line(466-d, 466-d, 466-d-l, 466-d, ciano); it.line(466-d, 466-d, 466-d, 466-d-l, ciano);
// --- 2. ANELLI ESTERNI ---
it.circle(cx, cy, 220, main_color); // Ora cambia colore con lo stato!
it.circle(cx, cy, 190, main_color);
// Decorazioni superiori (Segmenti)
for(int i = -30; i < 30; i+=3) {
float rad = (i - 90) * 0.01745;
int x1 = cx + cos(rad) * 200;
int y1 = cy + sin(rad) * 200;
int x2 = cx + cos(rad) * 215;
int y2 = cy + sin(rad) * 215;
it.line(x1, y1, x2, y2, main_color);
}
// --- 3. IL CUORE GEOMETRICO (Rombo) ---
float rot = 0;
if (s == 2) rot = t / 400.0; // Rotazione veloce se pensa
int size = 70;
if (s == 1 || s == 3) size = 70 + (sin(t/150.0)*8); // Battito
// Calcolo Rombo
int x_top = cx + cos(rot - 1.57) * size;
int y_top = cy + sin(rot - 1.57) * size;
int x_right = cx + cos(rot) * size;
int y_right = cy + sin(rot) * size;
int x_bot = cx + cos(rot + 1.57) * size;
int y_bot = cy + sin(rot + 1.57) * size;
int x_left = cx + cos(rot + 3.14) * size;
int y_left = cy + sin(rot + 3.14) * size;
it.line(x_top, y_top, x_right, y_right, main_color);
it.line(x_right, y_right, x_bot, y_bot, main_color);
it.line(x_bot, y_bot, x_left, y_left, main_color);
it.line(x_left, y_left, x_top, y_top, main_color);
// Quadrato interno fisso
it.rectangle(cx - 35, cy - 35, 70, 70, main_color);
// --- 4. EFFETTI SPECIALI PER STATO ---
// IDLE: Scansione Laser
if (s == 0) {
int scan_y = cy + (sin(t/1000.0) * 60); // Va su e giù sopra il rombo
it.line(cx - 50, scan_y, cx + 50, scan_y, ciano);
it.print(cx, 160, id(font_ui), ciano, TextAlign::CENTER, "SYSTEM READY");
it.print(cx, 300, id(font_ui), bianco, TextAlign::CENTER, "WAITING...");
}
// LISTENING: Onde verso l'interno
else if (s == 1) {
int r_wave = 100 - ((t % 1000) / 10); // Cerchio che si restringe
it.circle(cx, cy, r_wave, bianco);
it.print(cx, 160, id(font_ui), bianco, TextAlign::CENTER, "LISTENING");
}
// PARLO: Equalizzatore Grafico (Nuovo!)
else if (s == 3) {
it.print(cx, 160, id(font_ui), verde, TextAlign::CENTER, "JARVIS VOICE");
// Disegna 5 barre sotto al rombo
for (int i = -2; i <= 2; i++) {
// Altezza casuale "simulata" con seni sfasati
int h = 10 + (sin((t/100.0) + i) * 20) + 20;
int x_bar = cx + (i * 20);
int y_base = 320;
it.filled_rectangle(x_bar - 5, y_base - h, 10, h, verde);
}
}
// PENSO: Glitch Text
else if (s == 2) {
if ((t/200)%2 == 0) it.print(cx, 320, id(font_ui), viola, TextAlign::CENTER, "PROCESSING_");
else it.print(cx, 320, id(font_ui), viola, TextAlign::CENTER, "PROCESSING ");
}
// --- 5. DATA STREAM LATERALE (Fake Data) ---
// Una colonna di numeri che cambia a destra
it.print(400, 200, id(font_ui), ciano, TextAlign::CENTER, "DAT");
int val = (t / 100) % 99;
char buffer[10];
sprintf(buffer, "%02d", val);
it.print(400, 230, id(font_ui), bianco, TextAlign::CENTER, buffer);
sprintf(buffer, "%02d", (val * 3) % 99);
it.print(400, 260, id(font_ui), bianco, TextAlign::CENTER, buffer);
// --- ICONE STATICHE (Migliorate) ---
it.circle(100, 100, 15, ciano); // Mic
if (s == 1) it.filled_circle(100, 100, 10, bianco); // Accendi mic se ascolta
it.circle(366, 100, 15, verde); // Speaker
if (s == 3) it.filled_circle(366, 100, 10, verde); // Accendi speaker se parla
# =========================
# LUMINOSITÀ & FONT
# =========================
light:
- platform: monochromatic
id: display_backlight
name: "Backlight"
output: backlight_brightness
restore_mode: ALWAYS_ON
output:
- platform: template
id: backlight_brightness
type: float
write_action:
then:
- lambda: |-
id(disp1).set_brightness(state*255);
font:
- file: "gfonts://Roboto"
id: font_ui
size: 36
Nota tecnica: Usiamo update_interval: 33ms per garantire un refresh a 30Hz, fondamentale per l'effetto "film".
A questo punto annotatevi l'indirizzo IP che il vostro router assegna al Waveshare ed eventualmente prenotatelo (ip reservation) nel DHCP.
Se volete osservare il risultato della programmazione potete aprire un browser (chrome o quello che preferite) e digitare l'indirizzo ip del Waveshare nella barra degli indirizzi. Dovreste accedere alla pagine web del dispositivo dove c'è un selettore che vi permette di impostare manualmente i vari stati (e grafiche) del display.
Passo 2: Il "Cervello" (Configurazione ReSpeaker)
Qui le cose si fanno interessanti. Dobbiamo dire al ReSpeaker di inviare un comando HTTP al Waveshare ogni volta che cambia stato.
Il problema della sincronizzazione: Inizialmente usavamo i trigger del voice_assistant (on_tts_start, on_tts_end). Risultato? Un disastro. L'animazione "parlato" finiva prima che l'audio smettesse di uscire dalla cassa, oppure lampeggiava verde quando il sistema faceva un semplice "BIP" di wake-word.
La Soluzione: Monitorare direttamente lo stato fisico del media_player e sincronizzarlo con la fase dell'assistente vocale.
Ecco il blocco integrale per il ReSpeaker lite compreso il media_player "blindato" da inserire nel YAML del ReSpeaker:
Ricordatevi di personalizzare ssid e password della vostra rete WiFi e di intervenire sugli url http modificando l'indirizzo ip in base a quello che ha il Waveshare nella vostra rete.
substitutions:
# Phases of the Voice Assistant
# The voice assistant is ready to be triggered by a wake word
voice_assist_idle_phase_id: '1'
# The voice assistant is waiting for a voice command (after being triggered by the wake word)
voice_assist_waiting_for_command_phase_id: '2'
# The voice assistant is listening for a voice command
voice_assist_listening_for_command_phase_id: '3'
# The voice assistant is currently processing the command
voice_assist_thinking_phase_id: '4'
# The voice assistant is replying to the command
voice_assist_replying_phase_id: '5'
# The voice assistant is not ready
voice_assist_not_ready_phase_id: '10'
# The voice assistant encountered an error
voice_assist_error_phase_id: '11'
# Change this to true in case you ahve a hidden SSID at home.
hidden_ssid: "false"
# Substitutions for audio files
mute_switch_on_sound_file: https://github.com/esphome/home-assistant-voice-pe/raw/dev/sounds/mute_switch_on.flac
mute_switch_off_sound_file: https://github.com/esphome/home-assistant-voice-pe/raw/dev/sounds/mute_switch_off.flac
timer_finished_sound_file: https://github.com/esphome/home-assistant-voice-pe/raw/dev/sounds/timer_finished.flac
wake_word_triggered_sound_file: https://github.com/esphome/home-assistant-voice-pe/raw/dev/sounds/wake_word_triggered.flac
center_button_press_sound_file: https://github.com/esphome/home-assistant-voice-pe/raw/dev/sounds/center_button_press.flac
center_button_double_press_sound_file: https://github.com/esphome/home-assistant-voice-pe/raw/dev/sounds/center_button_double_press.flac
center_button_triple_press_sound_file: https://github.com/esphome/home-assistant-voice-pe/raw/dev/sounds/center_button_triple_press.flac
center_button_long_press_sound_file: https://github.com/esphome/home-assistant-voice-pe/raw/dev/sounds/center_button_long_press.flac
esphome:
name: resp_new
friendly_name: resp_new
project:
name: formatbce.Respeaker Lite Satellite
version: 2025.10.0
min_version: 2025.9.3
on_boot:
- priority: 375
then:
- sensor.template.publish:
id: next_timer
state: -1
# Run the script to refresh the LED status
- script.execute: control_leds
# If after 10 minutes, the device is still initializing (It did not yet connect to Home Assistant), turn off the init_in_progress variable and run the script to refresh the LED status
- delay: 10min
- if:
condition:
lambda: return id(init_in_progress);
then:
- lambda: id(init_in_progress) = false;
- script.execute: control_leds
- priority: -100
then:
- lambda: |-
id(alarm_time).publish_state(id(saved_alarm_time));
- lambda: |-
auto call = id(alarm_action).make_call();
call.set_option(id(saved_alarm_action));
call.perform();
- lambda: |-
setenv("TZ", id(saved_time_zone).c_str(), 1);
tzset();
on_shutdown:
then:
# Prevent loud noise on software restart
- lambda: id(respeaker).mute_speaker();
esp32:
board: esp32-s3-devkitc-1
cpu_frequency: 240MHz
variant: esp32s3
flash_size: 8MB
framework:
type: esp-idf
version: recommended
sdkconfig_options:
CONFIG_ESP32S3_DATA_CACHE_64KB: "y"
CONFIG_ESP32S3_DATA_CACHE_LINE_64B: "y"
CONFIG_ESP32S3_INSTRUCTION_CACHE_32KB: "y"
# Moves instructions and read only data from flash into PSRAM on boot.
# Both enabled allows instructions to execute while a flash operation is in progress without needing to be placed in IRAM.
# Considerably speeds up mWW at the cost of using more PSRAM.
CONFIG_SPIRAM_RODATA: "y"
CONFIG_SPIRAM_FETCH_INSTRUCTIONS: "y"
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" # TLS1.3 support isn't enabled by default in IDF 5.1.5
http_request:
useragent: esphome/jarvis-brain
timeout: 3s
wifi:
id: wifi_id
fast_connect: ${hidden_ssid}
on_connect:
- lambda: id(improv_ble_in_progress) = false;
- script.execute: control_leds
on_disconnect:
- script.execute: control_leds
ssid: "tuossid" #personalizza con il nome del tuo WiFi
password: "tuapasswordwifi" #personalizza con la password del tuo WiFi
network:
enable_ipv6: true
logger:
level: debug
initial_level: debug
logs:
sensor: WARN # avoids logging debug sensor updates
# hardware_uart: uart0 #Uncomment to see serial logs via USB connection. Comment out after debufgging - this line introduces noise on speaker...
api:
id: api_id
actions:
- action: start_va
then:
- voice_assistant.start
- action: stop_va
then:
- voice_assistant.stop
- action: set_alarm_time
variables:
alarm_time_hh_mm: string
then:
- lambda: |-
if (alarm_time_hh_mm.length() == 5 &&
isdigit(alarm_time_hh_mm[0]) && isdigit(alarm_time_hh_mm[1]) &&
isdigit(alarm_time_hh_mm[3]) && isdigit(alarm_time_hh_mm[4])) {
id(alarm_time).publish_state(alarm_time_hh_mm);
id(saved_alarm_time) = alarm_time_hh_mm;
}
- action: set_time_zone
variables:
posix_time_zone: string
then:
- lambda: |-
setenv("TZ", posix_time_zone.c_str(), 1);
tzset();
id(saved_time_zone) = posix_time_zone;
id(publish_current_time).execute();
on_client_connected:
- script.execute: control_leds
on_client_disconnected:
- script.execute: control_leds
encryption:
# Uncomment this, if you have problems with text-to-speech because of Home Assistant HTTPS internal URL
# http_request:
# verify_ssl: false
#ota:
# - platform: esphome
# id: ota_esphome
# password: !secret ota_password
i2c:
- id: internal_i2c
sda: GPIO5
scl: GPIO6
frequency: 400kHz
psram:
mode: octal
speed: 80MHz
globals:
- id: init_in_progress
type: bool
restore_value: no
initial_value: 'true'
# Global variable storing the state of ImprovBLE. Used to draw different LED animations
- id: improv_ble_in_progress
type: bool
restore_value: no
initial_value: 'false'
# Global variable tracking the phase of the voice assistant (defined above). Initialized to not_ready
- id: voice_assistant_phase
type: int
restore_value: no
initial_value: ${voice_assist_not_ready_phase_id}
- id: saved_alarm_time
type: std::string
restore_value: yes
initial_value: '"Unknown"'
- id: saved_time_zone
type: std::string
restore_value: yes
initial_value: '"UTC0"'
- id: saved_alarm_action
type: std::string
restore_value: yes
initial_value: '"Play sound"'
# Global variable storing the first active timer
- id: first_active_timer
type: voice_assistant::Timer
restore_value: no
# Global variable storing if a timer is active
- id: is_timer_active
type: bool
restore_value: no
# Global variable storing if a factory reset was requested. If it is set to true, the device will factory reset once the center button is released
- id: factory_reset_requested
type: bool
restore_value: no
initial_value: 'false'
# Time sync from Home Assistant
time:
- platform: homeassistant
id: homeassistant_time
on_time:
# Every 1 minute
- seconds: 0
minutes: /1
then:
- script.execute: check_alarm
on_time_sync:
- script.execute: publish_current_time
switch:
# Hardware speaker mute
- platform: template
id: speaker_mute_switch
name: Speaker mute
icon: mdi:volume-mute
internal: true
optimistic: true
turn_on_action:
- lambda: id(respeaker).mute_speaker();
turn_off_action:
- lambda: id(respeaker).unmute_speaker();
# stateless momentary mic mute switch
- platform: gpio
internal: true
pin:
number: GPIO4 # D3
inverted: true
id: mute_toggle
on_turn_on:
- delay: 300ms
- switch.turn_off: mute_toggle
# stateful user facing mic mute switch
- platform: template
id: mic_mute_switch
name: Mic mute
icon: mdi:microphone-off
optimistic: false
lambda: |-
if (id(mute_state).state) {
return true;
} else {
return false;
}
on_turn_on:
- if:
condition:
and:
- lambda: return !id(init_in_progress);
- switch.is_on: mute_sound
then:
- script.execute:
id: play_sound
priority: false
sound_file: !lambda return id(mute_switch_on_sound);
on_turn_off:
- if:
condition:
and:
- lambda: return !id(init_in_progress);
- switch.is_on: mute_sound
then:
- script.execute:
id: play_sound
priority: false
sound_file: !lambda return id(mute_switch_off_sound);
turn_on_action:
- if:
condition:
lambda: return !id(mute_state).state;
then:
- switch.turn_on: mute_toggle
turn_off_action:
- if:
condition:
lambda: return id(mute_state).state;
then:
- switch.turn_on: mute_toggle
# Button click Sounds Switch.
- platform: template
id: button_sound
name: Button click sounds
icon: "mdi:bullhorn"
entity_category: config
optimistic: true
restore_mode: RESTORE_DEFAULT_ON
# Mute Sound Switch.
- platform: template
id: mute_sound
name: Mute/unmute sound
icon: "mdi:bullhorn"
entity_category: config
optimistic: true
restore_mode: RESTORE_DEFAULT_ON
# Wake Word Sound Switch.
- platform: template
id: wake_sound
name: Wake sound
icon: "mdi:bullhorn"
entity_category: config
optimistic: true
restore_mode: RESTORE_DEFAULT_ON
# Internal switch to track when a timer is ringing on the device.
- platform: template
id: timer_ringing
optimistic: true
internal: true
restore_mode: ALWAYS_OFF
on_turn_off:
# Disable stop wake word
- micro_wake_word.disable_model: stop
- script.execute: disable_repeat
# Stop any current annoucement (ie: stop the timer ring mid playback)
- if:
condition:
media_player.is_announcing:
then:
media_player.stop:
announcement: true
# Set back ducking ratio to zero
- mixer_speaker.apply_ducking:
id: media_mixing_input
decibel_reduction: 0
duration: 1.0s
# Refresh the LED ring
- script.execute: control_leds
on_turn_on:
# Duck audio
- mixer_speaker.apply_ducking:
id: media_mixing_input
decibel_reduction: 20
duration: 0.0s
# Enable stop wake word
- micro_wake_word.enable_model: stop
# Ring timer
- script.execute: ring_timer
# Refresh LED
- script.execute: control_leds
# If 15 minutes have passed and the timer is still ringing, stop it.
- delay: 15min
- switch.turn_off: timer_ringing
# Defines if alarm is active
- platform: template
optimistic: true
restore_mode: RESTORE_DEFAULT_OFF
id: alarm_on
icon: mdi:bell-badge
name: "Alarm on"
on_turn_on:
- script.execute: control_leds
on_turn_off:
- script.execute: control_leds
binary_sensor:
# User Button. Used for many things (See on_multi_click)
- platform: gpio
id: user_button
pin:
number: GPIO3 # D2
inverted: true
name: "User button"
on_press:
- script.execute: control_leds
on_release:
- script.execute: control_leds
# If a factory reset is requested, factory reset on release
- if:
condition:
lambda: return id(factory_reset_requested);
then:
- button.press: factory_reset_button
on_multi_click:
# Simple Click:
# - Abort "things" in order
# - Timer
# - Announcements
# - Voice Assistant Pipeline run
# - Music
# - Starts the voice assistant if it is not yet running and if the device is not muted.
- timing:
- ON for at most 1s
- OFF for at least 0.25s
then:
- if:
condition:
lambda: return !id(init_in_progress);
then:
- if:
condition:
switch.is_on: timer_ringing
then:
- switch.turn_off: timer_ringing
else:
- if:
condition:
voice_assistant.is_running:
then:
- voice_assistant.stop:
else:
- if:
condition:
media_player.is_announcing:
then:
media_player.stop:
announcement: true
else:
- if:
condition:
media_player.is_playing:
then:
- media_player.pause:
else:
- if:
condition:
and:
- switch.is_off: mic_mute_switch
- not:
voice_assistant.is_running
then:
- if:
condition:
switch.is_on: button_sound
then:
- script.execute:
id: play_sound
priority: true
sound_file: !lambda return id(center_button_press_sound);
- delay: 300ms
- voice_assistant.start:
# Double Click
# . Exposed as an event entity. To be used in automations inside Home Assistant
- timing:
- ON for at most 1s
- OFF for at most 0.25s
- ON for at most 1s
- OFF for at least 0.25s
then:
- if:
condition:
lambda: return !id(init_in_progress);
then:
- if:
condition:
switch.is_on: button_sound
then:
- script.execute:
id: play_sound
priority: false
sound_file: !lambda return id(center_button_double_press_sound);
- event.trigger:
id: button_press_event
event_type: "double_press"
# Triple Click
# . Exposed as an event entity. To be used in automations inside Home Assistant
- timing:
- ON for at most 1s
- OFF for at most 0.25s
- ON for at most 1s
- OFF for at most 0.25s
- ON for at most 1s
- OFF for at least 0.25s
then:
- if:
condition:
lambda: return !id(init_in_progress);
then:
- if:
condition:
switch.is_on: button_sound
then:
- script.execute:
id: play_sound
priority: false
sound_file: !lambda return id(center_button_triple_press_sound);
- event.trigger:
id: button_press_event
event_type: "triple_press"
# Long Press
# . Exposed as an event entity. To be used in automations inside Home Assistant
- timing:
- ON for at least 1s
then:
- if:
condition:
lambda: return !id(init_in_progress);
then:
- if:
condition:
switch.is_on: button_sound
then:
- script.execute:
id: play_sound
priority: false
sound_file: !lambda return id(center_button_long_press_sound);
- light.turn_off: led_internal
- event.trigger:
id: button_press_event
event_type: "long_press"
# Factory Reset Warning
# . Audible and Visible warning.
- timing:
- ON for at least 10s
then:
- light.turn_on:
brightness: 100%
id: led_internal
effect: "Factory Reset Coming Up"
- script.execute:
id: play_sound
priority: true
sound_file: !lambda return id(factory_reset_initiated_sound);
- wait_until:
binary_sensor.is_off: user_button
- if:
condition:
lambda: return !id(factory_reset_requested);
then:
- light.turn_off: led_internal
- script.execute:
id: play_sound
priority: true
sound_file: !lambda return id(factory_reset_cancelled_sound);
# Factory Reset Confirmed.
# . Audible warning to prompt user to release the button
# . Set factory_reset_requested to true
- timing:
- ON for at least 22s
then:
- script.execute:
id: play_sound
priority: true
sound_file: !lambda return id(factory_reset_confirmed_sound);
- light.turn_on:
brightness: 100%
red: 100%
green: 0%
blue: 0%
id: led_internal
effect: "none"
- lambda: id(factory_reset_requested) = true;
light:
- platform: esp32_rmt_led_strip
id: led_internal
internal: true
rgb_order: GRB
pin: GPIO1
num_leds: 1
rmt_symbols: 192
chipset: ws2812
default_transition_length: 0s
effects:
- addressable_lambda:
name: "Fast Pulse"
update_interval: 10ms
lambda: |-
static float fraction = 0.0;
static float step = 0.05;
static bool increasing = true;
auto values = id(led_internal)->current_values;
Color color(values.get_red() * 255, values.get_green() * 255, values.get_blue() * 255);
it[0].set_rgb(color.red * fraction,
color.green * fraction,
color.blue * fraction);
fraction += (step * (increasing ? 1 : -1));
if (fraction > 1.0) {
fraction = 1.0;
increasing = !increasing;
} else if (fraction < 0.0) {
fraction = 0.0;
increasing = !increasing;
}
- addressable_lambda:
name: "Slow Pulse"
update_interval: 30ms
lambda: |-
static float fraction = 0.0;
static float step = 0.05;
static bool increasing = true;
auto values = id(led_internal)->current_values;
Color color(values.get_red() * 255, values.get_green() * 255, values.get_blue() * 255);
it[0].set_rgb(color.red * fraction,
color.green * fraction,
color.blue * fraction);
fraction += (step * (increasing ? 1 : -1));
if (fraction > 1.0) {
fraction = 1.0;
increasing = !increasing;
} else if (fraction < 0.0) {
fraction = 0.0;
increasing = !increasing;
}
# User facing LED.
# Exposed to be used by the user.
- platform: partition
id: led_respeaker_onboard
name: LED Respeaker onboard
entity_category: config
icon: "mdi:circle-outline"
default_transition_length: 0ms
restore_mode: RESTORE_DEFAULT_OFF
on_turn_off:
- script.execute: control_leds
initial_state:
color_mode: rgb
brightness: 100%
red: 9.4%
green: 73.3%
blue: 94.9%
segments:
- id: led_internal
from: 0
to: 0
effects:
- addressable_lambda:
name: "Fast Pulse"
update_interval: 10ms
lambda: |-
static float fraction = 0.0;
static float step = 0.05;
static bool increasing = true;
auto values = id(led_respeaker_onboard)->current_values;
Color color(values.get_red() * 255, values.get_green() * 255, values.get_blue() * 255);
it[0].set_rgb(color.red * fraction,
color.green * fraction,
color.blue * fraction);
fraction += (step * (increasing ? 1 : -1));
if (fraction > 1.0) {
fraction = 1.0;
increasing = !increasing;
} else if (fraction < 0.0) {
fraction = 0.0;
increasing = !increasing;
}
- addressable_lambda:
name: "Slow Pulse"
update_interval: 30ms
lambda: |-
static float fraction = 0.0;
static float step = 0.05;
static bool increasing = true;
auto values = id(led_respeaker_onboard)->current_values;
Color color(values.get_red() * 255, values.get_green() * 255, values.get_blue() * 255);
it[0].set_rgb(color.red * fraction,
color.green * fraction,
color.blue * fraction);
fraction += (step * (increasing ? 1 : -1));
if (fraction > 1.0) {
fraction = 1.0;
increasing = !increasing;
} else if (fraction < 0.0) {
fraction = 0.0;
increasing = !increasing;
}
sensor:
- platform: template
id: next_timer
name: "Next timer"
update_interval: never
disabled_by_default: true
device_class: duration
unit_of_measurement: s
icon: "mdi:timer"
text_sensor:
- platform: template
id: next_timer_name
name: "Next timer name"
icon: "mdi:timer"
disabled_by_default: true
- platform: template
name: "Alarm time"
id: alarm_time
icon: mdi:bell-ring
- platform: template
name: "Current device time"
id: current_time
icon: mdi:clock
- id: text_request
platform: template
name: "Text Request"
on_value:
lambda: |-
if(id(text_request).state.length()>245) {
std::string name = id(text_request).state.c_str();
std::string truncated = esphome::str_truncate(name.c_str(),245);
id(text_request).state = (truncated+" ...").c_str();
}
- id: text_response
platform: template
name: "Text Response"
on_value:
lambda: |-
if(id(text_response).state.length()>245) {
std::string name = id(text_response).state.c_str();
std::string truncated = esphome::str_truncate(name.c_str(),245);
id(text_response).state = (truncated+" ...").c_str();
}
- platform: template
id: voice_assistant_phase_sensor
name: "Voice Assistant Phase ID"
update_interval: never
event:
# Event entity exposed to the user to automate on complex center button presses.
# The simple press is not exposed as it is used to control the device itself.
- platform: template
id: button_press_event
name: "Button press"
icon: mdi:button-pointer
device_class: button
event_types:
- double_press
- triple_press
- long_press
script:
# Master script controlling the LEDs, based on different conditions : initialization in progress, wifi and api connected and voice assistant phase.
# For the sake of simplicity and re-usability, the script calls child scripts defined below.
# This script will be called every time one of these conditions is changing.
- id: control_leds
then:
- lambda: |
id(check_if_timers_active).execute();
if (id(is_timer_active)){
id(fetch_first_active_timer).execute();
}
if (id(improv_ble_in_progress)) {
id(control_leds_improv_ble_state).execute();
} else if (id(init_in_progress)) {
id(control_leds_init_state).execute();
} else if (!id(wifi_id).is_connected() || !id(api_id).is_connected()){
id(control_leds_no_ha_connection_state).execute();
} else if (id(user_button).state) {
id(control_leds_center_button_touched).execute();
} else if (id(timer_ringing).state) {
id(control_leds_timer_ringing).execute();
} else if (id(voice_assistant_phase) == ${voice_assist_waiting_for_command_phase_id}) {
id(control_leds_voice_assistant_waiting_for_command_phase).execute();
} else if (id(voice_assistant_phase) == ${voice_assist_listening_for_command_phase_id}) {
id(control_leds_voice_assistant_listening_for_command_phase).execute();
} else if (id(voice_assistant_phase) == ${voice_assist_thinking_phase_id}) {
id(control_leds_voice_assistant_thinking_phase).execute();
} else if (id(voice_assistant_phase) == ${voice_assist_replying_phase_id}) {
id(control_leds_voice_assistant_replying_phase).execute();
} else if (id(voice_assistant_phase) == ${voice_assist_error_phase_id}) {
id(control_leds_voice_assistant_error_phase).execute();
} else if (id(voice_assistant_phase) == ${voice_assist_not_ready_phase_id}) {
id(control_leds_voice_assistant_not_ready_phase).execute();
} else if (id(is_timer_active)) {
id(control_leds_timer_ticking).execute();
} else if (id(alarm_on).state && !id(led_respeaker_onboard).remote_values.is_on()) {
id(control_leds_alarm_active).execute();
} else if (id(voice_assistant_phase) == ${voice_assist_idle_phase_id}) {
id(control_leds_voice_assistant_idle_phase).execute();
}
# Script executed during Improv BLE
# Warm White slow pulse
- id: control_leds_improv_ble_state
then:
- light.turn_on:
brightness: !lambda return max( id(led_respeaker_onboard).current_values.get_brightness() , 0.2f );
red: 100%
green: 89%
blue: 71%
id: led_internal
effect: "Slow Pulse"
# Script executed during initialization
# Fast Blue pulse if Wifi is connected, Else slow blue pulse
- id: control_leds_init_state
then:
- if:
condition:
wifi.connected:
then:
- light.turn_on:
brightness: !lambda return max( id(led_respeaker_onboard).current_values.get_brightness() , 0.2f );
red: 9%
green: 73%
blue: 95%
id: led_internal
effect: "Fast Pulse"
else:
- light.turn_on:
brightness: !lambda return max( id(led_respeaker_onboard).current_values.get_brightness() , 0.2f );
red: 9%
green: 73%
blue: 95%
id: led_internal
effect: "Slow Pulse"
# Script executed when the device has no connection to Home Assistant
# Red slow pulse (This will be visible during HA updates for example)
- id: control_leds_no_ha_connection_state
then:
- light.turn_on:
brightness: !lambda return max( id(led_respeaker_onboard).current_values.get_brightness() , 0.2f );
red: 1
green: 0
blue: 0
id: led_internal
effect: "Slow Pulse"
# Script executed when the voice assistant is idle (waiting for a wake word)
# Nothing
- id: control_leds_voice_assistant_idle_phase
then:
- light.turn_off: led_internal
- if:
condition:
light.is_on: led_respeaker_onboard
then:
light.turn_on: led_respeaker_onboard
# Script executed when the voice assistant is waiting for a command (After the wake word)
# Slow purple pulse
- id: control_leds_voice_assistant_waiting_for_command_phase
then:
- light.turn_on:
brightness: !lambda return max( id(led_respeaker_onboard).current_values.get_brightness() , 0.2f );
red: 1
green: 0.2
blue: 1
id: led_internal
effect: "Slow Pulse"
# Script executed when the voice assistant is listening to a command
# Slow purple pulse
- id: control_leds_voice_assistant_listening_for_command_phase
then:
- light.turn_on:
brightness: !lambda return max( id(led_respeaker_onboard).current_values.get_brightness() , 0.2f );
red: 1
green: 0.2
blue: 1
id: led_internal
effect: "Slow Pulse"
# Script executed when the voice assistant is thinking to a command
# Fast purple pulse
- id: control_leds_voice_assistant_thinking_phase
then:
- light.turn_on:
brightness: !lambda return max( id(led_respeaker_onboard).current_values.get_brightness() , 0.2f );
red: 1
green: 0.2
blue: 1
id: led_internal
effect: "Fast Pulse"
# Script executed when the voice assistant is replying to a command
# Slow cyan pulse
- id: control_leds_voice_assistant_replying_phase
then:
- light.turn_on:
brightness: !lambda return max( id(led_respeaker_onboard).current_values.get_brightness() , 0.2f );
red: 0.2
green: 1
blue: 1
id: led_internal
effect: "Slow Pulse"
# Script executed when the voice assistant is in error
# Fast Red Pulse
- id: control_leds_voice_assistant_error_phase
then:
- light.turn_on:
brightness: !lambda return max( id(led_respeaker_onboard).current_values.get_brightness() , 0.2f );
red: 1
green: 0
blue: 0
id: led_internal
effect: "Fast Pulse"
# Script executed when the voice assistant is not ready
- id: control_leds_voice_assistant_not_ready_phase
then:
- light.turn_on:
brightness: !lambda return max( id(led_respeaker_onboard).current_values.get_brightness() , 0.2f );
red: 1
green: 0
blue: 0
id: led_internal
effect: "Slow Pulse"
# Script executed when the center button is touched
# The LED turns on blue
- id: control_leds_center_button_touched
then:
- light.turn_on:
brightness: !lambda return min ( max( id(led_respeaker_onboard).current_values.get_brightness() , 0.2f ) + 0.1f , 1.0f );
red: 0
green: 0
blue: 1
id: led_internal
effect: "None"
# Script executed when the timer is ringing, to control the LEDs
# The LED blinks green.
- id: control_leds_timer_ringing
then:
- light.turn_on:
brightness: !lambda return min ( max( id(led_respeaker_onboard).current_values.get_brightness() , 0.2f ) + 0.1f , 1.0f );
red: 0
green: 1
blue: 0
id: led_internal
effect: "Fast Pulse"
# Script executed when the timer is ticking, to control the LEDs
# Slow dim while pulse.
- id: control_leds_timer_ticking
then:
- light.turn_on:
brightness: !lambda return max( id(led_respeaker_onboard).current_values.get_brightness() , 0.2f );
red: 0.3
green: 0.3
blue: 0.3
id: led_internal
effect: "Slow Pulse"
# Script executed when the alarm is active
# The LED turns on dim green
- id: control_leds_alarm_active
then:
- light.turn_on:
brightness: !lambda return 0.3f;
red: 0
green: 1
blue: 0
id: led_internal
effect: "None"
# Script executed when the timer is ringing, to playback sounds.
- id: ring_timer
then:
- script.execute: enable_repeat_one
- script.execute:
id: play_sound
priority: true
sound_file: !lambda return id(timer_finished_sound);
# Script executed when the timer is ringing, to repeat the timer finished sound.
- id: enable_repeat_one
then:
# Turn on the repeat mode and pause for 500 ms between playlist items/repeats
- lambda: |-
id(external_media_player)
->make_call()
.set_command(media_player::MediaPlayerCommand::MEDIA_PLAYER_COMMAND_REPEAT_ONE)
.set_announcement(true)
.perform();
id(external_media_player)->set_playlist_delay_ms(speaker::AudioPipelineType::ANNOUNCEMENT, 500);
# Script execute when the timer is done ringing, to disable repeat mode.
- id: disable_repeat
then:
# Turn off the repeat mode and pause for 0 ms between playlist items/repeats
- lambda: |-
id(external_media_player)
->make_call()
.set_command(media_player::MediaPlayerCommand::MEDIA_PLAYER_COMMAND_REPEAT_OFF)
.set_announcement(true)
.perform();
id(external_media_player)->set_playlist_delay_ms(speaker::AudioPipelineType::ANNOUNCEMENT, 0);
# Script executed when we want to play sounds on the device.
- id: play_sound
parameters:
priority: bool
sound_file: "audio::AudioFile*"
then:
- lambda: |-
if (priority) {
id(external_media_player)
->make_call()
.set_command(media_player::MediaPlayerCommand::MEDIA_PLAYER_COMMAND_STOP)
.set_announcement(true)
.perform();
}
if ( (id(external_media_player).state != media_player::MediaPlayerState::MEDIA_PLAYER_STATE_ANNOUNCING ) || priority) {
id(external_media_player)
->play_file(sound_file, true, false);
}
# Script used to fetch the first active timer (Stored in global first_active_timer)
- id: fetch_first_active_timer
then:
- lambda: |
const auto timers = id(va).get_timers();
auto output_timer = timers.begin()->second;
for (auto &iterable_timer : timers) {
if (iterable_timer.second.is_active && iterable_timer.second.seconds_left <= output_timer.seconds_left) {
output_timer = iterable_timer.second;
}
}
id(first_active_timer) = output_timer;
# Script used to check if a timer is active (Stored in global is_timer_active)
- id: check_if_timers_active
then:
- lambda: |
const auto timers = id(va).get_timers();
bool output = false;
if (timers.size() > 0) {
for (auto &iterable_timer : timers) {
if(iterable_timer.second.is_active) {
output = true;
}
}
}
id(is_timer_active) = output;
# Script used activate the stop word if the TTS step is long.
# Why is this wrapped on a script?
# Becasue we want to stop the sequence if the TTS step is faster than that.
# This allows us to prevent having the deactivation of the stop word before its own activation.
- id: activate_stop_word_once
then:
- delay: 1s
# Enable stop wake word
- if:
condition:
switch.is_off: timer_ringing
then:
- micro_wake_word.enable_model: stop
- wait_until:
not:
media_player.is_announcing:
- if:
condition:
switch.is_off: timer_ringing
then:
- micro_wake_word.disable_model: stop
- id: check_alarm
then:
- lambda: |-
id(publish_current_time).execute();
// Check alarm
if (id(alarm_on).state && id(alarm_time).has_state()) {
// Get the stored alarm time from the sensor
auto set_alarm_time = id(alarm_time).state;
if (set_alarm_time.length() == 5 &&
isdigit(set_alarm_time[0]) && isdigit(set_alarm_time[1]) &&
isdigit(set_alarm_time[3]) && isdigit(set_alarm_time[4])) {
auto alarm_hour = std::stoi(set_alarm_time.substr(0, 2));
auto alarm_minute = std::stoi(set_alarm_time.substr(3, 2));
// Trigger action if current time matches alarm time
auto time_now = id(homeassistant_time).now();
if (time_now.hour == alarm_hour && time_now.minute == alarm_minute) {
auto action = id(alarm_action).state;
if (action == "Play sound") {
id(timer_ringing).turn_on();
} else if (action == "Send event") {
id(send_alarm_event).execute();
} else if (action == "Sound and event") {
id(timer_ringing).turn_on();
id(send_alarm_event).execute();
}
}
} else {
ESP_LOGW("alarm", "Incorrect alarm time setting");
}
}
- id: send_alarm_event
then:
- homeassistant.event:
event: esphome.alarm_ringing
- id: send_tts_uri_event
parameters:
tts_uri: string
then:
- homeassistant.event:
event: esphome.tts_uri
data:
uri: !lambda return tts_uri;
- id: send_stt_text_event
parameters:
stt_text: string
then:
- homeassistant.event:
event: esphome.stt_text
data:
text: !lambda return stt_text;
- id: publish_current_time
then:
- lambda: |-
// Publish current time
auto time_now = id(homeassistant_time).now();
id(current_time).publish_state(time_now.strftime("%H:%M"));
- id: set_phase
parameters:
phase: int
then:
- lambda: id(voice_assistant_phase) = phase;
- text_sensor.template.publish:
id: voice_assistant_phase_sensor
state: !lambda |-
char buf[8];
snprintf(buf, sizeof(buf), "%d", phase);
return std::string(buf);
- id: handle_screen_tap
then:
- if:
condition:
lambda: return !id(init_in_progress);
then:
- if:
condition:
switch.is_on: timer_ringing
then:
- switch.turn_off: timer_ringing
else:
- if:
condition:
voice_assistant.is_running:
then:
- voice_assistant.stop:
else:
- if:
condition:
media_player.is_announcing:
then:
media_player.stop:
announcement: true
else:
- if:
condition:
media_player.is_playing:
then:
- media_player.pause:
else:
- if:
condition:
and:
- switch.is_off: mic_mute_switch
- not:
voice_assistant.is_running
then:
- if:
condition:
switch.is_on: button_sound
then:
- script.execute:
id: play_sound
priority: true
sound_file: !lambda return id(center_button_press_sound);
- delay: 300ms
- voice_assistant.start:
i2s_audio:
- id: i2s_output
i2s_lrclk_pin:
number: GPIO7
allow_other_uses: true
i2s_bclk_pin:
number: GPIO8
allow_other_uses: true
i2s_mclk_pin:
number: GPIO9
allow_other_uses: true
- id: i2s_input
i2s_lrclk_pin:
number: GPIO7
allow_other_uses: true
i2s_bclk_pin:
number: GPIO8
allow_other_uses: true
i2s_mclk_pin:
number: GPIO9
allow_other_uses: true
microphone:
- platform: i2s_audio
id: i2s_mics
i2s_din_pin: GPIO44
adc_type: external
pdm: false
sample_rate: 48000
bits_per_sample: 32bit
i2s_mode: secondary
i2s_audio_id: i2s_input
channel: stereo
speaker:
# Hardware speaker output
- platform: i2s_audio
id: i2s_audio_speaker
sample_rate: 48000
i2s_mode: secondary
i2s_dout_pin: GPIO43
bits_per_sample: 32bit
i2s_audio_id: i2s_output
dac_type: external
channel: stereo
timeout: never
buffer_duration: 100ms
audio_dac: aic3204_dac
# Virtual speakers to combine the announcement and media streams together into one output
- platform: mixer
id: mixing_speaker
output_speaker: i2s_audio_speaker
num_channels: 2
source_speakers:
- id: announcement_mixing_input
timeout: never
- id: media_mixing_input
timeout: never
# Vritual speakers to resample each pipelines' audio, if necessary, as the mixer speaker requires the same sample rate
- platform: resampler
id: announcement_resampling_speaker
output_speaker: announcement_mixing_input
sample_rate: 48000
bits_per_sample: 16
- platform: resampler
id: media_resampling_speaker
output_speaker: media_mixing_input
sample_rate: 48000
bits_per_sample: 16
media_player:
- platform: speaker
id: external_media_player
name: None
internal: False
volume_increment: 0.05
volume_min: 0.4
volume_max: 0.85
announcement_pipeline:
speaker: announcement_resampling_speaker
format: FLAC
num_channels: 1
sample_rate: 48000
media_pipeline:
speaker: media_resampling_speaker
format: FLAC
num_channels: 2
sample_rate: 48000
on_announcement:
- mixer_speaker.apply_ducking:
id: media_mixing_input
decibel_reduction: 20
duration: 0.0s
on_state:
# 1. Ducking audio (Originale)
- if:
condition:
and:
- switch.is_off: timer_ringing
- not:
voice_assistant.is_running:
- not:
media_player.is_announcing:
then:
- mixer_speaker.apply_ducking:
id: media_mixing_input
decibel_reduction: 0
duration: 1.0s
# 2. INIZIO PARLATO (Verde) - CON FILTRO
- if:
condition:
and:
- media_player.is_announcing:
# TRUCCO: Controlla se siamo davvero nella fase di risposta!
- lambda: |-
return id(voice_assistant_phase) == ${voice_assist_replying_phase_id};
then:
- http_request.post: "http://192.168.1.99/number/jstate/set?value=3"
# 3. FINE PARLATO (Blu) - CON FILTRO
- if:
condition:
and:
- not:
media_player.is_announcing: # <--- CORRETTO QUI (A capo)
# TRUCCO: Torna blu solo se eravamo in fase di risposta
- lambda: |-
return id(voice_assistant_phase) == ${voice_assist_replying_phase_id};
then:
- http_request.post: "http://192.168.1.99/number/jstate/set?value=0"
files:
- id: center_button_press_sound
file: ${center_button_press_sound_file}
- id: center_button_double_press_sound
file: ${center_button_double_press_sound_file}
- id: center_button_triple_press_sound
file: ${center_button_triple_press_sound_file}
- id: center_button_long_press_sound
file: ${center_button_long_press_sound_file}
- id: factory_reset_initiated_sound
file: https://github.com/esphome/home-assistant-voice-pe/raw/dev/sounds/factory_reset_initiated.mp3
- id: factory_reset_cancelled_sound
file: https://github.com/esphome/home-assistant-voice-pe/raw/dev/sounds/factory_reset_cancelled.mp3
- id: factory_reset_confirmed_sound
file: https://github.com/esphome/home-assistant-voice-pe/raw/dev/sounds/factory_reset_confirmed.mp3
- id: mute_switch_on_sound
file: ${mute_switch_on_sound_file}
- id: mute_switch_off_sound
file: ${mute_switch_off_sound_file}
- id: timer_finished_sound
file: ${timer_finished_sound_file}
- id: wake_word_triggered_sound
file: ${wake_word_triggered_sound_file}
- id: error_cloud_expired
file: https://github.com/esphome/home-assistant-voice-pe/raw/dev/sounds/error_cloud_expired.mp3
respeaker_lite:
id: respeaker
reset_pin: GPIO2
mute_state:
internal: true
id: mute_state
firmware_version:
icon: mdi:application-cog
name: XMOS firmware version
internal: false
id: firmware_version
firmware:
url: https://github.com/formatBCE/Respeaker-Lite-ESPHome-integration/raw/refs/heads/main/respeaker_lite_i2s_dfu_firmware_48k_v1.1.0.bin
version: "1.1.0"
md5: 9297155d1bf3eb21a9d4db52a89ea0c6
on_begin:
- light.turn_on:
brightness: !lambda return max( id(led_respeaker_onboard).current_values.get_brightness() , 0.2f );
red: 50%
green: 50%
blue: 50%
id: led_internal
effect: "Slow Pulse"
on_end:
- light.turn_on:
brightness: !lambda return max( id(led_respeaker_onboard).current_values.get_brightness() , 0.2f );
red: 0%
green: 100%
blue: 0%
id: led_internal
effect: "Fast Pulse"
- delay: 3s
- light.turn_off:
id: led_internal
on_error:
- light.turn_on:
brightness: !lambda return max( id(led_respeaker_onboard).current_values.get_brightness() , 0.2f );
red: 100%
green: 0%
blue: 0%
id: led_internal
effect: "Fast Pulse"
- delay: 3s
- light.turn_off:
id: led_internal
external_components:
- source:
type: git
url: https://github.com/formatBCE/esphome
ref: respeaker_microphone
components:
- i2s_audio
refresh: 0s
- source:
type: git
url: https://github.com/formatBCE/Respeaker-Lite-ESPHome-integration
ref: main
components:
- respeaker_lite
refresh: 0s
audio_dac:
- platform: aic3204
id: aic3204_dac
i2c_id: internal_i2c
micro_wake_word:
id: mww
microphone:
microphone: i2s_mics
channels: 1
gain_factor: 4
stop_after_detection: false
models:
- model: https://github.com/kahrendt/microWakeWord/releases/download/okay_nabu_20241226.3/okay_nabu.json
# probability_cutoff: 0.8
id: okay_nabu
- model: https://raw.githubusercontent.com/formatBCE/Respeaker-Lite-ESPHome-integration/refs/heads/main/microwakeword/models/v2/kenobi.json
id: kenobi
- model: hey_jarvis
id: hey_jarvis
- model: hey_mycroft
id: hey_mycroft
- model: https://github.com/kahrendt/microWakeWord/releases/download/stop/stop.json
id: stop
internal: true
vad:
probability_cutoff: 0.05
on_wake_word_detected:
# If the wake word is detected when the device is muted (Possible with the software mute switch): Do nothing
- if:
condition:
switch.is_off: mic_mute_switch
then:
# If a timer is ringing: Stop it, do not start the voice assistant (We can stop timer from voice!)
- if:
condition:
switch.is_on: timer_ringing
then:
- switch.turn_off: timer_ringing
# Stop voice assistant if running
else:
- if:
condition:
voice_assistant.is_running:
then:
voice_assistant.stop:
# Stop any other media player announcement
else:
- if:
condition:
media_player.is_announcing:
then:
- media_player.stop:
announcement: true
# Start the voice assistant and play the wake sound, if enabled
else:
- if:
condition:
switch.is_on: wake_sound
then:
- script.execute:
id: play_sound
priority: true
sound_file: !lambda return id(wake_word_triggered_sound);
- delay: 300ms
- voice_assistant.start:
wake_word: !lambda return wake_word;
select:
- platform: template
name: "Wake word sensitivity"
optimistic: true
initial_option: Slightly 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(okay_nabu).set_probability_cutoff(217); // 0.85 -> 0.000 FAPH on DipCo (Manifest's default)
id(hey_jarvis).set_probability_cutoff(247); // 0.97 -> 0.563 FAPH on DipCo (Manifest's default)
id(hey_mycroft).set_probability_cutoff(253); // 0.99 -> 0.567 FAPH on DipCo
} else if (x == "Moderately sensitive") {
id(okay_nabu).set_probability_cutoff(176); // 0.69 -> 0.376 FAPH on DipCo
id(hey_jarvis).set_probability_cutoff(235); // 0.92 -> 0.939 FAPH on DipCo
id(hey_mycroft).set_probability_cutoff(242); // 0.95 -> 1.502 FAPH on DipCo (Manifest's default)
} else if (x == "Very sensitive") {
id(okay_nabu).set_probability_cutoff(143); // 0.56 -> 0.751 FAPH on DipCo
id(hey_jarvis).set_probability_cutoff(212); // 0.83 -> 1.502 FAPH on DipCo
id(hey_mycroft).set_probability_cutoff(237); // 0.93 -> 1.878 FAPH on DipCo
}
- platform: logger
id: logger_select
name: Logger Level
disabled_by_default: true
- platform: template
optimistic: true
name: "Alarm action"
id: alarm_action
icon: mdi:bell-plus
options:
- "Play sound"
- "Send event"
- "Sound and event"
initial_option: "Play sound"
on_value:
then:
- lambda: |-
id(saved_alarm_action) = x;
voice_assistant:
id: va
microphone:
microphone: i2s_mics
channels: 0
media_player: external_media_player
micro_wake_word: mww
use_wake_word: false
noise_suppression_level: 0
auto_gain: 0 dbfs
volume_multiplier: 1
on_client_connected:
- if:
condition:
- lambda: return id(init_in_progress);
- switch.is_on: mic_mute_switch
then:
- switch.turn_off: mic_mute_switch
- lambda: id(init_in_progress) = false;
- micro_wake_word.start:
- script.execute:
id: set_phase
phase: ${voice_assist_idle_phase_id}
- script.execute: control_leds
on_client_disconnected:
- voice_assistant.stop:
- script.execute:
id: set_phase
phase: ${voice_assist_not_ready_phase_id}
- script.execute: control_leds
on_error:
- http_request.post: "http://192.168.1.99/number/jstate/set?value=4" # RITIENE ERRORE (Rosso)
- if:
condition:
and:
- lambda: return !id(init_in_progress);
- lambda: return code != "duplicate_wake_up_detected";
- lambda: return code != "stt-no-text-recognized";
then:
- script.execute:
id: set_phase
phase: ${voice_assist_error_phase_id}
- script.execute: control_leds
- if:
condition:
- lambda: return code == "cloud-auth-failed";
then:
- script.execute:
id: play_sound
priority: true
sound_file: !lambda return id(error_cloud_expired);
on_start:
- mixer_speaker.apply_ducking:
id: media_mixing_input
decibel_reduction: 20
duration: 0.0s
on_listening:
- http_request.post: "http://192.168.1.99/number/jstate/set?value=1" # RITIENE ASCOLTO (Bianco/Rombo)
- script.execute:
id: set_phase
phase: ${voice_assist_waiting_for_command_phase_id}
- script.execute: control_leds
- text_sensor.template.publish:
id: text_request
state: "..."
- text_sensor.template.publish:
id: text_response
state: "..."
on_stt_vad_start:
- script.execute:
id: set_phase
phase: ${voice_assist_listening_for_command_phase_id}
- script.execute: control_leds
on_stt_vad_end:
- http_request.post: "http://192.168.1.99/number/jstate/set?value=2" # RITIENE PENSIERO (Viola)
- script.execute:
id: set_phase
phase: ${voice_assist_thinking_phase_id}
- script.execute: control_leds
on_intent_progress:
- if:
condition:
lambda: 'return !x.empty();'
then:
- script.execute:
id: set_phase
phase: ${voice_assist_replying_phase_id}
- script.execute: control_leds
- script.execute: activate_stop_word_once
on_tts_start:
# QUI ABBIAMO TOLTO IL VALUE=3 (Gestito ora da media_player)
- if:
condition:
lambda: 'return id(voice_assistant_phase) != ${voice_assist_replying_phase_id};'
then:
- text_sensor.template.publish:
id: text_response
state: !lambda return x;
- script.execute:
id: set_phase
phase: ${voice_assist_replying_phase_id}
- script.execute: control_leds
- script.execute: activate_stop_word_once
on_tts_end:
# QUI ABBIAMO TOLTO IL VALUE=0 (Gestito ora da media_player)
- script.execute:
id: send_tts_uri_event
tts_uri: !lambda 'return x;'
on_stt_end:
- script.execute:
id: send_stt_text_event
stt_text: !lambda 'return x;'
- text_sensor.template.publish:
id: text_request
state: !lambda return x;
on_end:
- wait_until:
not:
voice_assistant.is_running:
- mixer_speaker.apply_ducking:
id: media_mixing_input
decibel_reduction: 0
duration: 1.0s
- if:
condition:
lambda: return id(voice_assistant_phase) == ${voice_assist_error_phase_id};
then:
- delay: 1s
- script.execute:
id: set_phase
phase: ${voice_assist_idle_phase_id}
- script.execute: control_leds
on_timer_finished:
- switch.turn_on: timer_ringing
- lambda: |
id(next_timer).publish_state(-1);
id(next_timer_name).publish_state("-");
on_timer_started:
- script.execute: control_leds
- lambda: |
id(next_timer).publish_state(id(first_active_timer).seconds_left);
id(next_timer_name).publish_state(id(first_active_timer).name);
on_timer_cancelled:
- script.execute: control_leds
- lambda: |
id(next_timer).publish_state(id(first_active_timer).seconds_left);
id(next_timer_name).publish_state(id(first_active_timer).name);
on_timer_updated:
- script.execute: control_leds
- lambda: |
id(next_timer).publish_state(id(first_active_timer).seconds_left);
id(next_timer_name).publish_state(id(first_active_timer).name);
on_timer_tick:
- script.execute: control_leds
- lambda: |
int seconds_left = id(first_active_timer).seconds_left;
if (std::abs(seconds_left) % 5 == 0) {
id(next_timer).publish_state(seconds_left);
}
button:
- platform: factory_reset
id: factory_reset_button
name: "Factory Reset"
entity_category: diagnostic
internal: true
- platform: restart
id: restart_button
name: "Restart"
entity_category: config
disabled_by_default: true
icon: "mdi:restart"
- platform: template
name: "Virtual Touch"
id: virtual_touch
on_press:
- script.execute: handle_screen_tap
debug:
update_interval: 5s
Nel voice_assistant, invece, gestiamo solo gli stati di "Input":
on_listening -> Invia 1 (Ascolto)
on_stt_vad_end -> Invia 2 (Elaborazione/Pensiero)
on_error -> Invia 4 (Errore)
Il Risultato Finale
Dopo circa 15 minuti di compilazione (sì, l'ESP32-S3 ci mette un po' a compilare le librerie audio, approfittatene per un caffè), il risultato è magico.
Dici "Hey Jarvis": Il Waveshare passa istantaneamente da una scansione blu a un occhio bianco pulsante.
Chiedi "Che tempo fa?": Mentre Home Assistant elabora (e il ReSpeaker invia l'audio al server), il display mostra un anello viola che ruota vorticosamente (stato Pensiero).
La Risposta: Appena la voce inizia a uscire dallo speaker, sul display appare un equalizzatore verde sincronizzato. Niente lag, niente falsi positivi.
Silenzio: Tutto torna blu, pronto per il prossimo comando.
Conclusione
Abbiamo imparato che l'hardware "all-in-one" è comodo, ma dividere i compiti è professionale. Il ReSpeaker Lite offre un audio cristallino che il Waveshare si sogna, e il Waveshare offre una grafica che il ReSpeaker non può gestire. Insieme, collegati da un filo invisibile di richieste HTTP, creano l'assistente DIY definitivo.
E voi? Siete pronti a costruire il vostro reattore Arc casalingo? Fateci sapere nei commenti se avete provato questa configurazione!
Restiamo Connessi
Se volete ricevere una notifica istantanea ogni volta che pubblico una nuova guida o un aggiornamento su queste integrazioni AI (e altri interessanti contenuti), vi invito a iscrivervi al mio Canale Telegram ufficiale:
Iscriviti al Canale Telegram di Vincenzo Caputo
Guarda il video su MissingTech
Volete vedere il Waveshare in azione? Nel video dedicato sul mio canale YouTube MissingTech, vi mostrerò il test dal vivo e una prova completa con la stupenda grafica di questo display touch!
Produrre e aggiornare contenuti su vincenzocaputo.com richiede molto tempo e lavoro. Se il contenuto che hai appena letto è di tuo gradimento e vuoi supportarmi, clicca uno dei link qui sotto per fare una donazione.