From fdebf041967549866f438431284c0690bec3d6da Mon Sep 17 00:00:00 2001
From: Kevin Ahrendt <kevin.ahrendt@nabucasa.com>
Date: Wed, 23 Oct 2024 13:25:31 -0400
Subject: [PATCH] [voice_assistant] Bugfix: Fix crash on start (#7662)

---
 .../voice_assistant/voice_assistant.cpp       | 54 +++++++++++--------
 .../voice_assistant/voice_assistant.h         |  6 +--
 2 files changed, 34 insertions(+), 26 deletions(-)

diff --git a/esphome/components/voice_assistant/voice_assistant.cpp b/esphome/components/voice_assistant/voice_assistant.cpp
index 0b53e74ba3..6f164f69d3 100644
--- a/esphome/components/voice_assistant/voice_assistant.cpp
+++ b/esphome/components/voice_assistant/voice_assistant.cpp
@@ -433,16 +433,18 @@ void VoiceAssistant::loop() {
 
 #ifdef USE_SPEAKER
 void VoiceAssistant::write_speaker_() {
-  if (this->speaker_buffer_size_ > 0) {
-    size_t write_chunk = std::min<size_t>(this->speaker_buffer_size_, 4 * 1024);
-    size_t written = this->speaker_->play(this->speaker_buffer_, write_chunk);
-    if (written > 0) {
-      memmove(this->speaker_buffer_, this->speaker_buffer_ + written, this->speaker_buffer_size_ - written);
-      this->speaker_buffer_size_ -= written;
-      this->speaker_buffer_index_ -= written;
-      this->set_timeout("speaker-timeout", 5000, [this]() { this->speaker_->stop(); });
-    } else {
-      ESP_LOGV(TAG, "Speaker buffer full, trying again next loop");
+  if ((this->speaker_ != nullptr) && (this->speaker_buffer_ != nullptr)) {
+    if (this->speaker_buffer_size_ > 0) {
+      size_t write_chunk = std::min<size_t>(this->speaker_buffer_size_, 4 * 1024);
+      size_t written = this->speaker_->play(this->speaker_buffer_, write_chunk);
+      if (written > 0) {
+        memmove(this->speaker_buffer_, this->speaker_buffer_ + written, this->speaker_buffer_size_ - written);
+        this->speaker_buffer_size_ -= written;
+        this->speaker_buffer_index_ -= written;
+        this->set_timeout("speaker-timeout", 5000, [this]() { this->speaker_->stop(); });
+      } else {
+        ESP_LOGV(TAG, "Speaker buffer full, trying again next loop");
+      }
     }
   }
 }
@@ -772,16 +774,20 @@ void VoiceAssistant::on_event(const api::VoiceAssistantEventResponse &msg) {
     }
     case api::enums::VOICE_ASSISTANT_TTS_STREAM_START: {
 #ifdef USE_SPEAKER
-      this->wait_for_stream_end_ = true;
-      ESP_LOGD(TAG, "TTS stream start");
-      this->defer([this] { this->tts_stream_start_trigger_->trigger(); });
+      if (this->speaker_ != nullptr) {
+        this->wait_for_stream_end_ = true;
+        ESP_LOGD(TAG, "TTS stream start");
+        this->defer([this] { this->tts_stream_start_trigger_->trigger(); });
+      }
 #endif
       break;
     }
     case api::enums::VOICE_ASSISTANT_TTS_STREAM_END: {
 #ifdef USE_SPEAKER
-      this->stream_ended_ = true;
-      ESP_LOGD(TAG, "TTS stream end");
+      if (this->speaker_ != nullptr) {
+        this->stream_ended_ = true;
+        ESP_LOGD(TAG, "TTS stream end");
+      }
 #endif
       break;
     }
@@ -802,14 +808,16 @@ void VoiceAssistant::on_event(const api::VoiceAssistantEventResponse &msg) {
 
 void VoiceAssistant::on_audio(const api::VoiceAssistantAudio &msg) {
 #ifdef USE_SPEAKER  // We should never get to this function if there is no speaker anyway
-  if (this->speaker_buffer_index_ + msg.data.length() < SPEAKER_BUFFER_SIZE) {
-    memcpy(this->speaker_buffer_ + this->speaker_buffer_index_, msg.data.data(), msg.data.length());
-    this->speaker_buffer_index_ += msg.data.length();
-    this->speaker_buffer_size_ += msg.data.length();
-    this->speaker_bytes_received_ += msg.data.length();
-    ESP_LOGV(TAG, "Received audio: %u bytes from API", msg.data.length());
-  } else {
-    ESP_LOGE(TAG, "Cannot receive audio, buffer is full");
+  if ((this->speaker_ != nullptr) && (this->speaker_buffer_ != nullptr)) {
+    if (this->speaker_buffer_index_ + msg.data.length() < SPEAKER_BUFFER_SIZE) {
+      memcpy(this->speaker_buffer_ + this->speaker_buffer_index_, msg.data.data(), msg.data.length());
+      this->speaker_buffer_index_ += msg.data.length();
+      this->speaker_buffer_size_ += msg.data.length();
+      this->speaker_bytes_received_ += msg.data.length();
+      ESP_LOGV(TAG, "Received audio: %u bytes from API", msg.data.length());
+    } else {
+      ESP_LOGE(TAG, "Cannot receive audio, buffer is full");
+    }
   }
 #endif
 }
diff --git a/esphome/components/voice_assistant/voice_assistant.h b/esphome/components/voice_assistant/voice_assistant.h
index 870e2acdaa..0016d3157c 100644
--- a/esphome/components/voice_assistant/voice_assistant.h
+++ b/esphome/components/voice_assistant/voice_assistant.h
@@ -250,7 +250,7 @@ class VoiceAssistant : public Component {
 #ifdef USE_SPEAKER
   void write_speaker_();
   speaker::Speaker *speaker_{nullptr};
-  uint8_t *speaker_buffer_;
+  uint8_t *speaker_buffer_{nullptr};
   size_t speaker_buffer_index_{0};
   size_t speaker_buffer_size_{0};
   size_t speaker_bytes_received_{0};
@@ -282,8 +282,8 @@ class VoiceAssistant : public Component {
   float volume_multiplier_;
   uint32_t conversation_timeout_;
 
-  uint8_t *send_buffer_;
-  int16_t *input_buffer_;
+  uint8_t *send_buffer_{nullptr};
+  int16_t *input_buffer_{nullptr};
 
   bool continuous_{false};
   bool silence_detection_;