Upstream/esphome
1
0
Fork 0
mirror of https://github.com/esphome/esphome.git synced 2024-11-21 11:37:27 +01:00
Code Issues Projects Releases Wiki Activity

[voice_assistant] Bugfix: Fix crash on start (#7662)

Browse Source
This commit is contained in:
Kevin Ahrendt 2024-10-23 13:25:31 -04:00 committed by GitHub
parent dd8d25e43f
commit fdebf04196
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 34 additions and 26 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

54
esphome/components/voice_assistant/voice_assistant.cpp
View File

@ -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
}

6
esphome/components/voice_assistant/voice_assistant.h
View File

@ -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_;