Speaker support (#4743)

CODEOWNERS

@@ -115,6 +115,7 @@ esphome/components/i2c/* @esphome/core
 esphome/components/i2s_audio/* @jesserockz
 esphome/components/i2s_audio/media_player/* @jesserockz
 esphome/components/i2s_audio/microphone/* @jesserockz
+esphome/components/i2s_audio/speaker/* @jesserockz
 esphome/components/ili9xxx/* @nielsnl68
 esphome/components/improv_base/* @esphome/core
 esphome/components/improv_serial/* @esphome/core
@@ -245,6 +246,7 @@ esphome/components/smt100/* @piechade
 esphome/components/sn74hc165/* @jesserockz
 esphome/components/socket/* @esphome/core
 esphome/components/sonoff_d1/* @anatoly-savchenkov
+esphome/components/speaker/* @jesserockz
 esphome/components/spi/* @esphome/core
 esphome/components/sprinkler/* @kbx81
 esphome/components/sps30/* @martgras

esphome/components/api/api_connection.cpp

@@ -996,7 +996,7 @@ DeviceInfoResponse APIConnection::device_info(const DeviceInfoRequest &msg) {
                                      : bluetooth_proxy::PASSIVE_ONLY_VERSION;
 #endif
 #ifdef USE_VOICE_ASSISTANT
-  resp.voice_assistant_version = 1;
+  resp.voice_assistant_version = voice_assistant::global_voice_assistant->get_version();
 #endif
   return resp;
 }

esphome/components/i2s_audio/i2s_audio.h

@@ -19,15 +19,6 @@ class I2SAudioComponent : public Component {
  public:
   void setup() override;
 
-  void register_audio_in(I2SAudioIn *in) {
-    this->audio_in_ = in;
-    in->set_parent(this);
-  }
-  void register_audio_out(I2SAudioOut *out) {
-    this->audio_out_ = out;
-    out->set_parent(this);
-  }
-
   i2s_pin_config_t get_pin_config() const {
     return {
         .mck_io_num = I2S_PIN_NO_CHANGE,
@@ -50,9 +41,6 @@ class I2SAudioComponent : public Component {
  protected:
   Mutex lock_;
 
-  I2SAudioIn *audio_in_{nullptr};
-  I2SAudioOut *audio_out_{nullptr};
-
   uint8_t bclk_pin_;
   uint8_t lrclk_pin_;
   i2s_port_t port_{};

esphome/components/i2s_audio/media_player/__init__.py

@@ -84,8 +84,7 @@ async def to_code(config):
     await cg.register_component(var, config)
     await media_player.register_media_player(var, config)
 
-    parent = await cg.get_variable(config[CONF_I2S_AUDIO_ID])
-    cg.add(parent.register_audio_out(var))
+    await cg.register_parented(var, config[CONF_I2S_AUDIO_ID])
 
     if config[CONF_DAC_TYPE] == "internal":
         cg.add(var.set_internal_dac_mode(config[CONF_MODE]))

esphome/components/i2s_audio/microphone/__init__.py

@@ -33,8 +33,7 @@ async def to_code(config):
     var = cg.new_Pvariable(config[CONF_ID])
     await cg.register_component(var, config)
 
-    parent = await cg.get_variable(config[CONF_I2S_AUDIO_ID])
-    cg.add(parent.register_audio_in(var))
+    await cg.register_parented(var, config[CONF_I2S_AUDIO_ID])
 
     cg.add(var.set_din_pin(config[CONF_I2S_DIN_PIN]))
 

esphome/components/i2s_audio/speaker/__init__.py

@@ -0,0 +1,87 @@
+import esphome.codegen as cg
+import esphome.config_validation as cv
+from esphome import pins
+from esphome.const import CONF_ID, CONF_MODE
+from esphome.components import esp32, speaker
+
+from .. import (
+    CONF_I2S_AUDIO_ID,
+    CONF_I2S_DOUT_PIN,
+    I2SAudioComponent,
+    I2SAudioOut,
+    i2s_audio_ns,
+)
+
+CODEOWNERS = ["@jesserockz"]
+DEPENDENCIES = ["i2s_audio"]
+
+I2SAudioSpeaker = i2s_audio_ns.class_(
+    "I2SAudioSpeaker", cg.Component, speaker.Speaker, I2SAudioOut
+)
+
+i2s_dac_mode_t = cg.global_ns.enum("i2s_dac_mode_t")
+
+CONF_MUTE_PIN = "mute_pin"
+CONF_DAC_TYPE = "dac_type"
+
+INTERNAL_DAC_OPTIONS = {
+    "left": i2s_dac_mode_t.I2S_DAC_CHANNEL_LEFT_EN,
+    "right": i2s_dac_mode_t.I2S_DAC_CHANNEL_RIGHT_EN,
+    "stereo": i2s_dac_mode_t.I2S_DAC_CHANNEL_BOTH_EN,
+}
+
+EXTERNAL_DAC_OPTIONS = ["mono", "stereo"]
+
+NO_INTERNAL_DAC_VARIANTS = [esp32.const.VARIANT_ESP32S2]
+
+
+def validate_esp32_variant(config):
+    if config[CONF_DAC_TYPE] != "internal":
+        return config
+    variant = esp32.get_esp32_variant()
+    if variant in NO_INTERNAL_DAC_VARIANTS:
+        raise cv.Invalid(f"{variant} does not have an internal DAC")
+    return config
+
+
+CONFIG_SCHEMA = cv.All(
+    cv.typed_schema(
+        {
+            "internal": speaker.SPEAKER_SCHEMA.extend(
+                {
+                    cv.GenerateID(): cv.declare_id(I2SAudioSpeaker),
+                    cv.GenerateID(CONF_I2S_AUDIO_ID): cv.use_id(I2SAudioComponent),
+                    cv.Required(CONF_MODE): cv.enum(INTERNAL_DAC_OPTIONS, lower=True),
+                }
+            ).extend(cv.COMPONENT_SCHEMA),
+            "external": speaker.SPEAKER_SCHEMA.extend(
+                {
+                    cv.GenerateID(): cv.declare_id(I2SAudioSpeaker),
+                    cv.GenerateID(CONF_I2S_AUDIO_ID): cv.use_id(I2SAudioComponent),
+                    cv.Required(
+                        CONF_I2S_DOUT_PIN
+                    ): pins.internal_gpio_output_pin_number,
+                    cv.Optional(CONF_MODE, default="mono"): cv.one_of(
+                        *EXTERNAL_DAC_OPTIONS, lower=True
+                    ),
+                }
+            ).extend(cv.COMPONENT_SCHEMA),
+        },
+        key=CONF_DAC_TYPE,
+    ),
+    validate_esp32_variant,
+)
+
+
+async def to_code(config):
+    var = cg.new_Pvariable(config[CONF_ID])
+    await cg.register_component(var, config)
+    await speaker.register_speaker(var, config)
+
+    await cg.register_parented(var, config[CONF_I2S_AUDIO_ID])
+
+    if config[CONF_DAC_TYPE] == "internal":
+        cg.add(var.set_internal_dac_mode(config[CONF_MODE]))
+    else:
+        cg.add(var.set_dout_pin(config[CONF_I2S_DOUT_PIN]))
+        cg.add(var.set_external_dac_channels(2 if config[CONF_MODE] == "stereo" else 1))

esphome/components/i2s_audio/speaker/i2s_audio_speaker.cpp

@@ -0,0 +1,208 @@
+#include "i2s_audio_speaker.h"
+
+#ifdef USE_ESP32
+
+#include <driver/i2s.h>
+
+#include "esphome/core/application.h"
+#include "esphome/core/hal.h"
+#include "esphome/core/log.h"
+
+namespace esphome {
+namespace i2s_audio {
+
+static const size_t BUFFER_COUNT = 10;
+
+static const char *const TAG = "i2s_audio.speaker";
+
+void I2SAudioSpeaker::setup() {
+  ESP_LOGCONFIG(TAG, "Setting up I2S Audio Speaker...");
+
+  this->buffer_queue_ = xQueueCreate(BUFFER_COUNT, sizeof(DataEvent));
+  this->event_queue_ = xQueueCreate(20, sizeof(TaskEvent));
+}
+
+void I2SAudioSpeaker::start() { this->state_ = speaker::STATE_STARTING; }
+void I2SAudioSpeaker::start_() {
+  if (!this->parent_->try_lock()) {
+    return;  // Waiting for another i2s component to return lock
+  }
+  this->state_ = speaker::STATE_RUNNING;
+
+  xTaskCreate(I2SAudioSpeaker::player_task, "speaker_task", 8192, (void *) this, 0, &this->player_task_handle_);
+}
+
+void I2SAudioSpeaker::player_task(void *params) {
+  I2SAudioSpeaker *this_speaker = (I2SAudioSpeaker *) params;
+
+  TaskEvent event;
+  event.type = TaskEventType::STARTING;
+  xQueueSend(this_speaker->event_queue_, &event, portMAX_DELAY);
+
+  i2s_driver_config_t config = {
+      .mode = (i2s_mode_t) (I2S_MODE_MASTER | I2S_MODE_TX),
+      .sample_rate = 16000,
+      .bits_per_sample = I2S_BITS_PER_SAMPLE_16BIT,
+      .channel_format = I2S_CHANNEL_FMT_RIGHT_LEFT,
+      .communication_format = I2S_COMM_FORMAT_STAND_I2S,
+      .intr_alloc_flags = ESP_INTR_FLAG_LEVEL1,
+      .dma_buf_count = 8,
+      .dma_buf_len = 1024,
+      .use_apll = false,
+      .tx_desc_auto_clear = true,
+      .fixed_mclk = I2S_PIN_NO_CHANGE,
+      .mclk_multiple = I2S_MCLK_MULTIPLE_DEFAULT,
+      .bits_per_chan = I2S_BITS_PER_CHAN_DEFAULT,
+  };
+#if SOC_I2S_SUPPORTS_DAC
+  if (this_speaker->internal_dac_mode_ != I2S_DAC_CHANNEL_DISABLE) {
+    config.mode = (i2s_mode_t) (config.mode | I2S_MODE_DAC_BUILT_IN);
+  }
+#endif
+
+  i2s_driver_install(this_speaker->parent_->get_port(), &config, 0, nullptr);
+
+#if SOC_I2S_SUPPORTS_DAC
+  if (this_speaker->internal_dac_mode_ == I2S_DAC_CHANNEL_DISABLE) {
+#endif
+    i2s_pin_config_t pin_config = this_speaker->parent_->get_pin_config();
+    pin_config.data_out_num = this_speaker->dout_pin_;
+
+    i2s_set_pin(this_speaker->parent_->get_port(), &pin_config);
+#if SOC_I2S_SUPPORTS_DAC
+  } else {
+    i2s_set_dac_mode(this_speaker->internal_dac_mode_);
+  }
+#endif
+
+  DataEvent data_event;
+
+  event.type = TaskEventType::STARTED;
+  xQueueSend(this_speaker->event_queue_, &event, portMAX_DELAY);
+
+  int16_t buffer[BUFFER_SIZE / 2];
+
+  while (true) {
+    if (xQueueReceive(this_speaker->buffer_queue_, &data_event, 100 / portTICK_PERIOD_MS) != pdTRUE) {
+      break;  // End of audio from main thread
+    }
+    if (data_event.stop) {
+      // Stop signal from main thread
+      while (xQueueReceive(this_speaker->buffer_queue_, &data_event, 0) == pdTRUE) {
+        // Flush queue
+      }
+      break;
+    }
+    size_t bytes_written;
+
+    memmove(buffer, data_event.data, data_event.len);
+    size_t remaining = data_event.len / 2;
+    size_t current = 0;
+
+    while (remaining > 0) {
+      uint32_t sample = (buffer[current] << 16) | (buffer[current] & 0xFFFF);
+
+      esp_err_t err = i2s_write(this_speaker->parent_->get_port(), &sample, sizeof(sample), &bytes_written,
+                                (100 / portTICK_PERIOD_MS));
+      if (err != ESP_OK) {
+        event = {.type = TaskEventType::WARNING, .err = err};
+        xQueueSend(this_speaker->event_queue_, &event, portMAX_DELAY);
+        continue;
+      }
+      remaining--;
+      current++;
+    }
+
+    event.type = TaskEventType::PLAYING;
+    xQueueSend(this_speaker->event_queue_, &event, portMAX_DELAY);
+  }
+
+  i2s_zero_dma_buffer(this_speaker->parent_->get_port());
+
+  event.type = TaskEventType::STOPPING;
+  xQueueSend(this_speaker->event_queue_, &event, portMAX_DELAY);
+
+  i2s_stop(this_speaker->parent_->get_port());
+  i2s_driver_uninstall(this_speaker->parent_->get_port());
+
+  event.type = TaskEventType::STOPPED;
+  xQueueSend(this_speaker->event_queue_, &event, portMAX_DELAY);
+
+  while (true) {
+    delay(10);
+  }
+}
+
+void I2SAudioSpeaker::stop() {
+  if (this->state_ == speaker::STATE_STOPPED)
+    return;
+  this->state_ = speaker::STATE_STOPPING;
+  DataEvent data;
+  data.stop = true;
+  xQueueSendToFront(this->buffer_queue_, &data, portMAX_DELAY);
+}
+
+void I2SAudioSpeaker::watch_() {
+  TaskEvent event;
+  if (xQueueReceive(this->event_queue_, &event, 0) == pdTRUE) {
+    switch (event.type) {
+      case TaskEventType::STARTING:
+      case TaskEventType::STARTED:
+      case TaskEventType::STOPPING:
+        break;
+      case TaskEventType::PLAYING:
+        this->status_clear_warning();
+        break;
+      case TaskEventType::STOPPED:
+        this->parent_->unlock();
+        this->state_ = speaker::STATE_STOPPED;
+        vTaskDelete(this->player_task_handle_);
+        this->player_task_handle_ = nullptr;
+        break;
+      case TaskEventType::WARNING:
+        ESP_LOGW(TAG, "Error writing to I2S: %s", esp_err_to_name(event.err));
+        this->status_set_warning();
+        break;
+    }
+  }
+}
+
+void I2SAudioSpeaker::loop() {
+  switch (this->state_) {
+    case speaker::STATE_STARTING:
+      this->start_();
+      break;
+    case speaker::STATE_RUNNING:
+      this->watch_();
+      break;
+    case speaker::STATE_STOPPING:
+    case speaker::STATE_STOPPED:
+      break;
+  }
+}
+
+bool I2SAudioSpeaker::play(const uint8_t *data, size_t length) {
+  if (this->state_ != speaker::STATE_RUNNING && this->state_ != speaker::STATE_STARTING) {
+    this->start();
+  }
+  size_t remaining = length;
+  size_t index = 0;
+  while (remaining > 0) {
+    DataEvent event;
+    event.stop = false;
+    size_t to_send_length = std::min(remaining, BUFFER_SIZE);
+    event.len = to_send_length;
+    memcpy(event.data, data + index, to_send_length);
+    if (xQueueSend(this->buffer_queue_, &event, 100 / portTICK_PERIOD_MS) == pdTRUE) {
+      remaining -= to_send_length;
+      index += to_send_length;
+    }
+    App.feed_wdt();
+  }
+  return true;
+}
+
+}  // namespace i2s_audio
+}  // namespace esphome
+
+#endif  // USE_ESP32

esphome/components/i2s_audio/speaker/i2s_audio_speaker.h

@@ -0,0 +1,81 @@
+#pragma once
+
+#ifdef USE_ESP32
+
+#include "../i2s_audio.h"
+
+#include <driver/i2s.h>
+#include <freertos/FreeRTOS.h>
+#include <freertos/queue.h>
+
+#include "esphome/components/speaker/speaker.h"
+#include "esphome/core/component.h"
+#include "esphome/core/gpio.h"
+#include "esphome/core/helpers.h"
+
+namespace esphome {
+namespace i2s_audio {
+
+static const size_t BUFFER_SIZE = 1024;
+
+enum class TaskEventType : uint8_t {
+  STARTING = 0,
+  STARTED,
+  PLAYING,
+  STOPPING,
+  STOPPED,
+  WARNING = 255,
+};
+
+struct TaskEvent {
+  TaskEventType type;
+  esp_err_t err;
+};
+
+struct DataEvent {
+  bool stop;
+  size_t len;
+  uint8_t data[BUFFER_SIZE];
+};
+
+class I2SAudioSpeaker : public Component, public speaker::Speaker, public I2SAudioOut {
+ public:
+  float get_setup_priority() const override { return esphome::setup_priority::LATE; }
+
+  void setup() override;
+  void loop() override;
+
+  void set_dout_pin(uint8_t pin) { this->dout_pin_ = pin; }
+#if SOC_I2S_SUPPORTS_DAC
+  void set_internal_dac_mode(i2s_dac_mode_t mode) { this->internal_dac_mode_ = mode; }
+#endif
+  void set_external_dac_channels(uint8_t channels) { this->external_dac_channels_ = channels; }
+
+  void start();
+  void stop() override;
+
+  bool play(const uint8_t *data, size_t length) override;
+
+ protected:
+  void start_();
+  // void stop_();
+  void watch_();
+
+  static void player_task(void *params);
+
+  TaskHandle_t player_task_handle_{nullptr};
+  QueueHandle_t buffer_queue_;
+  QueueHandle_t event_queue_;
+
+  uint8_t dout_pin_{0};
+
+#if SOC_I2S_SUPPORTS_DAC
+  i2s_dac_mode_t internal_dac_mode_{I2S_DAC_CHANNEL_DISABLE};
+#endif
+  uint8_t external_dac_channels_;
+};
+
+}  // namespace i2s_audio
+}  // namespace esphome
+
+#endif  // USE_ESP32

esphome/components/speaker/__init__.py

@@ -0,0 +1,87 @@
+from esphome import automation
+import esphome.config_validation as cv
+import esphome.codegen as cg
+
+from esphome.automation import maybe_simple_id
+from esphome.const import CONF_ID, CONF_DATA
+from esphome.core import CORE
+from esphome.coroutine import coroutine_with_priority
+
+
+CODEOWNERS = ["@jesserockz"]
+
+IS_PLATFORM_COMPONENT = True
+
+speaker_ns = cg.esphome_ns.namespace("speaker")
+
+Speaker = speaker_ns.class_("Speaker")
+
+PlayAction = speaker_ns.class_(
+    "PlayAction", automation.Action, cg.Parented.template(Speaker)
+)
+StopAction = speaker_ns.class_(
+    "StopAction", automation.Action, cg.Parented.template(Speaker)
+)
+
+IsPlayingCondition = speaker_ns.class_("IsPlayingCondition", automation.Condition)
+
+
+async def setup_speaker_core_(var, config):
+    pass
+
+
+async def register_speaker(var, config):
+    if not CORE.has_id(config[CONF_ID]):
+        var = cg.Pvariable(config[CONF_ID], var)
+    await setup_speaker_core_(var, config)
+
+
+SPEAKER_SCHEMA = cv.Schema({})
+
+
+SPEAKER_AUTOMATION_SCHEMA = maybe_simple_id({cv.GenerateID(): cv.use_id(Speaker)})
+
+
+async def speaker_action(config, action_id, template_arg, args):
+    var = cg.new_Pvariable(action_id, template_arg)
+    await cg.register_parented(var, config[CONF_ID])
+    return var
+
+
+@automation.register_action(
+    "speaker.play",
+    PlayAction,
+    cv.maybe_simple_value(
+        {
+            cv.GenerateID(): cv.use_id(Speaker),
+            cv.Required(CONF_DATA): cv.templatable(cv.ensure_list(cv.hex_uint8_t)),
+        },
+        key=CONF_DATA,
+    ),
+)
+async def speaker_play_action(config, action_id, template_arg, args):
+    var = cg.new_Pvariable(action_id, template_arg)
+    await cg.register_parented(var, config[CONF_ID])
+    data = config[CONF_DATA]
+
+    if cg.is_template(data):
+        templ = await cg.templatable(data, args, cg.std_vector.template(cg.uint8))
+        cg.add(var.set_data_template(templ))
+    else:
+        cg.add(var.set_data_static(data))
+    return var
+
+
+automation.register_action("speaker.stop", StopAction, SPEAKER_AUTOMATION_SCHEMA)(
+    speaker_action
+)
+
+automation.register_condition(
+    "speaker.is_playing", IsPlayingCondition, SPEAKER_AUTOMATION_SCHEMA
+)(speaker_action)
+
+
+@coroutine_with_priority(100.0)
+async def to_code(config):
+    cg.add_global(speaker_ns.using)
+    cg.add_define("USE_SPEAKER")

esphome/components/speaker/automation.h

@@ -0,0 +1,48 @@
+#pragma once
+
+#include "esphome/core/automation.h"
+#include "speaker.h"
+
+#include <vector>
+
+namespace esphome {
+namespace speaker {
+
+template<typename... Ts> class PlayAction : public Action<Ts...>, public Parented<Speaker> {
+ public:
+  void set_data_template(std::function<std::vector<uint8_t>(Ts...)> func) {
+    this->data_func_ = func;
+    this->static_ = false;
+  }
+  void set_data_static(const std::vector<uint8_t> &data) {
+    this->data_static_ = data;
+    this->static_ = true;
+  }
+
+  void play(Ts... x) override {
+    if (this->static_) {
+      this->parent_->play(this->data_static_);
+    } else {
+      auto val = this->data_func_(x...);
+      this->parent_->play(val);
+    }
+  }
+
+ protected:
+  bool static_{false};
+  std::function<std::vector<uint8_t>(Ts...)> data_func_{};
+  std::vector<uint8_t> data_static_{};
+};
+
+template<typename... Ts> class StopAction : public Action<Ts...>, public Parented<Speaker> {
+ public:
+  void play(Ts... x) override { this->parent_->stop(); }
+};
+
+template<typename... Ts> class IsPlayingCondition : public Condition<Ts...>, public Parented<Speaker> {
+ public:
+  bool check(Ts... x) override { return this->parent_->is_running(); }
+};
+
+}  // namespace speaker
+}  // namespace esphome

esphome/components/speaker/speaker.h

@@ -0,0 +1,27 @@
+#pragma once
+
+namespace esphome {
+namespace speaker {
+
+enum State : uint8_t {
+  STATE_STOPPED = 0,
+  STATE_STARTING,
+  STATE_RUNNING,
+  STATE_STOPPING,
+};
+
+class Speaker {
+ public:
+  virtual bool play(const uint8_t *data, size_t length) = 0;
+  virtual bool play(const std::vector<uint8_t> &data) { return this->play(data.data(), data.size()); }
+
+  virtual void stop() = 0;
+
+  bool is_running() const { return this->state_ == STATE_RUNNING; }
+
+ protected:
+  State state_{STATE_STOPPED};
+};
+
+}  // namespace speaker
+}  // namespace esphome

esphome/components/voice_assistant/__init__.py

@@ -1,10 +1,10 @@
 import esphome.config_validation as cv
 import esphome.codegen as cg
 
-from esphome.const import CONF_ID, CONF_MICROPHONE
+from esphome.const import CONF_ID, CONF_MICROPHONE, CONF_SPEAKER
 from esphome import automation
 from esphome.automation import register_action
-from esphome.components import microphone
+from esphome.components import microphone, speaker
 
 AUTO_LOAD = ["socket"]
 DEPENDENCIES = ["api", "microphone"]
@@ -34,6 +34,7 @@ CONFIG_SCHEMA = cv.Schema(
     {
         cv.GenerateID(): cv.declare_id(VoiceAssistant),
         cv.GenerateID(CONF_MICROPHONE): cv.use_id(microphone.Microphone),
+        cv.Optional(CONF_SPEAKER): cv.use_id(speaker.Speaker),
         cv.Optional(CONF_ON_START): automation.validate_automation(single=True),
         cv.Optional(CONF_ON_STT_END): automation.validate_automation(single=True),
         cv.Optional(CONF_ON_TTS_START): automation.validate_automation(single=True),
@@ -51,6 +52,10 @@ async def to_code(config):
     mic = await cg.get_variable(config[CONF_MICROPHONE])
     cg.add(var.set_microphone(mic))
 
+    if CONF_SPEAKER in config:
+        spkr = await cg.get_variable(config[CONF_SPEAKER])
+        cg.add(var.set_speaker(spkr))
+
     if CONF_ON_START in config:
         await automation.build_automation(
             var.get_start_trigger(), [], config[CONF_ON_START]

esphome/components/voice_assistant/voice_assistant.cpp

@@ -2,6 +2,8 @@
 
 #include "esphome/core/log.h"
 
+#include <cstdio>
+
 namespace esphome {
 namespace voice_assistant {
 
@@ -33,6 +35,27 @@ void VoiceAssistant::setup() {
     return;
   }
 
+#ifdef USE_SPEAKER
+  if (this->speaker_ != nullptr) {
+    struct sockaddr_storage server;
+
+    socklen_t sl = socket::set_sockaddr_any((struct sockaddr *) &server, sizeof(server), 6055);
+    if (sl == 0) {
+      ESP_LOGW(TAG, "Socket unable to set sockaddr: errno %d", errno);
+      this->mark_failed();
+      return;
+    }
+    server.ss_family = AF_INET;
+
+    err = socket_->bind((struct sockaddr *) &server, sizeof(server));
+    if (err != 0) {
+      ESP_LOGW(TAG, "Socket unable to bind: errno %d", errno);
+      this->mark_failed();
+      return;
+    }
+  }
+#endif
+
   this->mic_->add_data_callback([this](const std::vector<uint8_t> &data) {
     if (!this->running_) {
       return;
@@ -41,6 +64,21 @@ void VoiceAssistant::setup() {
   });
 }
 
+void VoiceAssistant::loop() {
+#ifdef USE_SPEAKER
+  if (this->speaker_ == nullptr) {
+    return;
+  }
+
+  uint8_t buf[1024];
+  auto len = this->socket_->read(buf, sizeof(buf));
+  if (len == -1) {
+    return;
+  }
+  this->speaker_->play(buf, len);
+#endif
+}
+
 void VoiceAssistant::start(struct sockaddr_storage *addr, uint16_t port) {
   ESP_LOGD(TAG, "Starting...");
 

esphome/components/voice_assistant/voice_assistant.h

@@ -2,23 +2,44 @@
 
 #include "esphome/core/automation.h"
 #include "esphome/core/component.h"
+#include "esphome/core/defines.h"
 #include "esphome/core/helpers.h"
 
 #include "esphome/components/api/api_pb2.h"
 #include "esphome/components/api/api_server.h"
 #include "esphome/components/microphone/microphone.h"
+#ifdef USE_SPEAKER
+#include "esphome/components/speaker/speaker.h"
+#endif
 #include "esphome/components/socket/socket.h"
 
 namespace esphome {
 namespace voice_assistant {
 
+// Version 1: Initial version
+// Version 2: Adds raw speaker support
+static const uint32_t INITIAL_VERSION = 1;
+static const uint32_t SPEAKER_SUPPORT = 2;
+
 class VoiceAssistant : public Component {
  public:
   void setup() override;
+  void loop() override;
   float get_setup_priority() const override;
   void start(struct sockaddr_storage *addr, uint16_t port);
 
   void set_microphone(microphone::Microphone *mic) { this->mic_ = mic; }
+#ifdef USE_SPEAKER
+  void set_speaker(speaker::Speaker *speaker) { this->speaker_ = speaker; }
+#endif
+
+  uint32_t get_version() const {
+#ifdef USE_SPEAKER
+    if (this->speaker_ != nullptr)
+      return SPEAKER_SUPPORT;
+#endif
+    return INITIAL_VERSION;
+  }
 
   void request_start();
   void signal_stop();
@@ -44,6 +65,9 @@ class VoiceAssistant : public Component {
   Trigger<std::string, std::string> *error_trigger_ = new Trigger<std::string, std::string>();
 
   microphone::Microphone *mic_{nullptr};
+#ifdef USE_SPEAKER
+  speaker::Speaker *speaker_{nullptr};
+#endif
 
   bool running_{false};
 };

esphome/const.py

@@ -659,6 +659,7 @@ CONF_SLEEP_WHEN_DONE = "sleep_when_done"
 CONF_SONY = "sony"
 CONF_SOURCE = "source"
 CONF_SOURCE_ID = "source_id"
+CONF_SPEAKER = "speaker"
 CONF_SPEED = "speed"
 CONF_SPEED_COMMAND_TOPIC = "speed_command_topic"
 CONF_SPEED_COUNT = "speed_count"

esphome/core/defines.h

@@ -73,6 +73,8 @@
 #define USE_WIFI_11KV_SUPPORT
 #define USE_BLUETOOTH_PROXY
 #define USE_VOICE_ASSISTANT
+#define USE_MICROPHONE
+#define USE_SPEAKER
 
 #ifdef USE_ARDUINO
 #define USE_ARDUINO_VERSION_CODE VERSION_CODE(2, 0, 5)

tests/test4.yaml

@@ -703,6 +703,13 @@ microphone:
     id: mic_id
     i2s_din_pin: GPIO23
 
+speaker:
+  - platform: i2s_audio
+    id: speaker_id
+    dac_type: external
+    i2s_dout_pin: GPIO25
+    mode: mono
+
 
 voice_assistant:
   microphone: mic_id