ESPHome
substitutions:
# HA scripts and entities
pipeline: "select.voice_assist_pipeline"
finished_speaking: "select.voice_finished_speaking_detection"
script_pipeline: "script.voice_assist_loop_pipeline"
script_finished_speaking: "script.voice_assist_loop_finished"
script_emoji: "script.voice_assist_emoji"
script_eesti: "script.voice_assist_eesti"
script_process_emoji: "script.voice_assist_emoji_process"
emojigpt: "input_text.emojigpt"
# display
default_url: "http://homeassistant.local:8123/local/yeet.png"
local_image_size: '128x32'
local_image_format: 'TRANSPARENT_BINARY'
local_image_use_transparency: 'true'
BOARD_TFT_HEIGHT: '32'
BOARD_TFT_WIDTH: '128'
globals:
- id: choice
type: std::string
restore_value: no
initial_value: '"Katsu"'
- id: voice_error_code
type: std::string
restore_value: no
initial_value: '"No error"'
- id: choices
type: std::vector<std::string>
restore_value: no
initial_value: '{"Küsida võib kõike","Katsu mind ja ma värisen","Ma olen ChatGPT"}'
switch:
- platform: template
name: Wake Word
id: use_wake_word
optimistic: true
restore_mode: RESTORE_DEFAULT_ON
icon: mdi:microphone-settings
on_turn_on:
- lambda: |-
const char* icon = "mdi:microphone";
id(use_wake_word).set_icon(icon);
id(${device_name}_voice_assistant).set_use_wake_word(true);
- if:
condition:
not:
- voice_assistant.is_running:
then:
- voice_assistant.start_continuous:
- lambda: |-
id(${device_name}_top).turn_on().set_rgb(0.00, 0.74, 0.83).set_effect("Alexircle").perform(); // cyan
on_turn_off:
- voice_assistant.stop:
- lambda: |-
const char* icon = "mdi:microphone-off";
id(use_wake_word).set_icon(icon);
id(${device_name}_voice_assistant).set_use_wake_word(false);
- delay: ${phi}
voice_assistant:
id: ${device_name}_voice_assistant
microphone: ${device_name}_microphone
media_player: ${device_name}_media_player
use_wake_word: true
noise_suppression_level: 2
auto_gain: 31dBFS
volume_multiplier: 2.0
on_client_connected:
- lambda: |-
ESP_LOGD("voice_assistant", "on_client_connected" );
- if:
condition:
switch.is_on: use_wake_word
then:
lambda: |-
id(start_voice_assistant)->execute(true);
on_client_disconnected:
- lambda: |-
ESP_LOGD("voice_assistant", "on_client_disconnected" );
- if:
condition:
switch.is_on: use_wake_word
then:
- voice_assistant.stop:
on_wake_word_detected:
then:
- lambda: |-
ESP_LOGD("voice_assistant", "on_wake_word_detected" );
id(start_voice_assistant)->execute(false);
on_listening:
then: # CHUT UP AND LISTEN
- if:
condition:
- media_player.is_playing: ${device_name}_media_player
then:
- media_player.stop: ${device_name}_media_player
- lambda: |-
ESP_LOGI("voice_assistant", "on_listening" );
- lambda: |-
id(${device_name}_top).turn_on().set_rgb(0.00, 0.74, 0.83).set_effect("Alexircle").perform(); // cyan
on_start:
then:
- lambda: |-
ESP_LOGD("voice_assistant", "on_start" );
id(backlight).turn_on().set_brightness(1.0).perform();
- if:
condition:
- media_player.is_playing: ${device_name}_media_player
then:
- media_player.stop: ${device_name}_media_player
on_stt_end:
then:
- lambda: |-
ESP_LOGD("voice_assistant", "on_stt_end");
- lambda: |-
std::string str = x.c_str();
str = str.substr(0, 255);
id(${device_name}_stt).publish_state(str);
- display.page.show: stt
- component.update: ssd1306_display
on_tts_start:
then:
- lambda: |-
ESP_LOGD("voice_assistant", "on_tts_start" );
- lambda: |-
std::string str = x.c_str();
str = str.substr(0, 255);
if ( "EmojiGPT" == id(assistant).state ){
id(chatgpt_emoji).publish_state(str);
}
else{
id(${device_name}_tts).publish_state(str);
}
- homeassistant.service:
service: ${script_process_emoji}
- if:
condition:
- lambda: |-
return id(assistant).state == "EmojiGPT";
then:
- display.page.show: chatgpt_picture
else:
- display.page.show: tts
- component.update: ssd1306_display
- lambda: |-
id(${device_name}_top).turn_on().set_rgb(0.30, 0.69, 0.31).perform(); // green
on_end:
then:
- lambda: |-
ESP_LOGD("voice_assistant", "on_end");
id(${device_name}_top).turn_off().set_brightness(0.0).perform();
- wait_until:
- not:
- media_player.is_playing: ${device_name}_media_player
- lambda: |-
id(play_sound)->execute("voice/alexa_listened");
on_error:
then:
- lambda: |-
auto error = code.c_str();
id(voice_error_code) = message.c_str();
ESP_LOGE("voice_assistant", "on_error %s %s", error, message.c_str() );
if (strcasecmp(error, "stt-no-text-recognized") == 0) { // orange
id(${device_name}_bot).turn_on().set_rgb(1.0, 0.59, 0.0).set_effect("Flash").perform();
}
else if (strcasecmp(error, "not-connected") == 0) { // red
id(${device_name}_bot).turn_on().set_rgb(0.95, 0.21, 0.21).set_effect("Flash").perform();
}
else if (strcasecmp(error, "stt-stream-failed") == 0) { // amber
id(${device_name}_bot).turn_on().set_rgb(1.00, 0.76, 0.03).set_effect("Flash").perform();
}
else if (strcasecmp(error, "pipeline-timeout") == 0) { // yellow
id(${device_name}_bot).turn_on().set_rgb(1.00, 0.92, 0.23).set_effect("Flash").perform();
}
else { // purple
id(${device_name}_bot).turn_on().set_rgb(0.61, 0.14, 0.69).set_effect("Flash").perform();
}
- display.page.show: error
- component.update: ssd1306_display
- lambda: |-
id(play_sound)->execute("voice/alexa_error");
- lambda: |-
if (code == "wake-provider-missing" || code == "wake-engine-missing") {
ESP_LOGW("voice_assistant", "%s", code.c_str() );
id(use_wake_word).turn_off();
}
- delay: $phi_x2
text_sensor:
- platform: template
lambda: |-
auto moji = id(emoji_response).state;
return moji;
name: Emoji Response
entity_category: diagnostic
icon: mdi:emoticon-cool-outline
- platform: homeassistant
entity_id: ${emojigpt}
id: emoji_response
name: Emoji Response
on_value:
then:
- display.page.show: chatgpt_picture
- component.update: ssd1306_display
- platform: homeassistant
entity_id: ${pipeline}
id: assistant
name: (Voice Assist pipeline)
internal: true
- platform: homeassistant
entity_id: ${finished_speaking}
name: (Voice Finished speaking detection)
internal: true
- platform: template
id: ${device_name}_tts
name: TTS
icon: mdi:speaker-message
update_interval: never
- platform: template
id: ${device_name}_stt
name: STT
icon: mdi:microphone-message
update_interval: never
- platform: template
name: Emoji
id: chatgpt_emoji
internal: true
update_interval: never
on_value:
- display.page.show: chatgpt_picture
- component.update: ssd1306_display
display: # https://esphome.io/components/display/ssd1306.html
- platform: ssd1306_i2c
id: ssd1306_display
model: "SSD1306 128x32"
update_interval: never
invert: false
rotation: 180°
contrast: 100%
flip_x: true
flip_y: true
address: 0x3C
pages: # screen size: 128x32
- id: idle
lambda: |-
it.fill(COLOR_OFF);
int w = it.get_width();
int h = it.get_height();
int x = w/2;
int y = h/2;
int size = 8;
static int offset = size;
std::string text = id(choice);
int text_width = text.length() * size; // text size / 2 ~= width;
if (text_width > it.get_width()) {
it.printf(x - offset, y, id(text_medium), COLOR_ON, TextAlign::CENTER_LEFT, "%s", text.c_str());
offset += size * 2;
if (offset > text_width) {
offset = size;
}
} else {
it.printf(x, y, id(text_large), COLOR_ON, TextAlign::CENTER, "%s", text.c_str());
}
- id: stt
lambda: |-
it.fill(COLOR_OFF);
int w = it.get_width();
int h = it.get_height();
int x = w/2;
int y = h/2;
int size = 8;
static int offset = size;
std::string text = id(${device_name}_stt).state;
int text_width = text.length() * size; // text size / 2 ~= width;
int wordCount = std::count(text.begin(), text.end(), ' ') + 1;
if (text_width > it.get_width()) {
it.printf(0 - offset, y, id(text_medium), COLOR_ON, TextAlign::CENTER_LEFT, "%s", text.c_str());
offset += size * 2;
if (offset > text_width) {
offset = size;
}
} else {
it.printf(x, y, id(text_large), COLOR_ON, TextAlign::CENTER, "%s", text.c_str());
}
- id: tts
lambda: |-
it.fill(COLOR_OFF);
int w = it.get_width();
int h = it.get_height();
int x = w/2;
int y = h/2;
int size = 8;
static int offset = size;
std::string text = id(${device_name}_tts).state;
int text_width = text.length() * size; // text size / 2 ~= width;
int wordCount = std::count(text.begin(), text.end(), ' ') + 1;
if (text_width > it.get_width()) {
it.printf(0 - offset, y, id(text_medium), COLOR_ON, TextAlign::CENTER_LEFT, "%s", text.c_str());
offset += size * 2;
if (offset > text_width) {
offset = 0;
}
} else {
it.printf(x, y, id(text_large), COLOR_ON, TextAlign::CENTER, "%s", text.c_str());
}
- id: error
lambda: |-
it.fill(COLOR_ON);
int w = it.get_width();
int h = it.get_height();
int x = w/2;
int y = h/2;
int size = 8;
static int offset = size;
std::string text = id(voice_error_code);
int text_width = text.length() * size; // text size / 2 ~= width;
// Count the number of spaces in the text
int wordCount = std::count(text.begin(), text.end(), ' ') + 1;
if (text_width > it.get_width()) {
it.printf(x - offset, y, id(text_medium), COLOR_OFF, TextAlign::CENTER_LEFT, "%s", text.c_str());
offset += size * 2;
if (offset > text_width) {
offset = size;
}
} else {
it.printf(x, y, id(text_large), COLOR_OFF, TextAlign::CENTER, "%s", text.c_str());
}
- id: chatgpt_picture
lambda: |-
it.fill(COLOR_OFF);
int w = it.get_width();
int h = it.get_height();
int x = w/2;
int y = h/2;
int size = 8;
std::string text = id(emoji_response).state;
it.printf(x, y, id(emojifont), COLOR_ON, TextAlign::CENTER, "%s", text.c_str());
- id: page_online_image
lambda: |-
it.fill(COLOR_OFF);
it.image(0, 0, id(example_image));
on_page_change:
- lambda: |-
auto page = id(ssd1306_display).get_active_page();
if ( to != id(page_online_image)) { // from == id(page_online_image) &&
ESP_LOGD("online_image","released for not being on page");
id(example_image)->release();
}
if (to == id(page_online_image) ) {
auto defaulturl = to_string("${default_url}");
if ( id(online_image_source).state == "" && id(online_image_source).state != defaulturl ){
auto call = id(online_image_source).make_call();
call.set_value(defaulturl.c_str());
call.perform();
}
}
if (to == id(idle) ) {
id(randomize).execute();
}
external_components:
- source: github://pr#4710
components: [ online_image ]
refresh: 1d
online_image:
- url: ${default_url}
id: example_image
type: $local_image_format
use_transparency: $local_image_use_transparency
useragent: ${friendly_name}
update_interval: never
on_download_finished:
- lambda: |-
id(ssd1306_display).show_page(id(page_online_image));
id(ssd1306_display).update();
on_error:
- lambda: |-
ESP_LOGW("online_image","could not download");
script:
- id: pipeline_next
mode: single
then:
- script.execute: interaction
- homeassistant.service:
service: ${script_pipeline}
- id: finished_speaking_next
mode: single
then:
- script.execute: interaction
- homeassistant.service:
service: ${script_finished_speaking}
- id: start_voice_assistant
parameters:
continuous: bool
then:
- if: # CHUT UP AND LISTEN
condition:
- media_player.is_playing: ${device_name}_media_player
then:
- media_player.stop: ${device_name}_media_player
- if: # toggle voice assistant
condition: voice_assistant.is_running
then:
- voice_assistant.stop:
else:
- if:
condition:
lambda: 'return continuous;'
then:
- voice_assistant.start_continuous:
else:
- voice_assistant.start:
silence_detection: true
HA script:
alias: Voice Emoji Process
sequence:
- service: conversation.process
data:
agent_id: xyz
text: |-
{%- if input is defined -%}
{{input}}
{%- else -%}
{{ states('sensor.voice_tts') }}
{%- endif -%}
language: et
response_variable: conversation_response
- service: input_text.set_value
data:
value: "{{ conversation_response.response.speech.plain.speech }}"
target:
entity_id: input_text.emojigpt
Voice Assistant prompt:
Translate this message into emojis Only use these emojis in response[ … ] Don’t use more than 5 emojis. Do not use words.
STT:
Emoji:
STT:
ERR:
Source image:
My HA is whacked, can’t upgrade from and it keeps crashing
Installed version 2024.2.5
Latest version 2024.3.0
Also my openAI needs funds, too much testing 
