Upstream/esphome
1
0
Fork 0
mirror of https://github.com/esphome/esphome.git synced 2025-02-22 02:42:02 +01:00
Code Issues Projects Releases Wiki Activity

[voice_assistant] Add announce support (#8232)

Browse Source
This commit is contained in:
Kevin Ahrendt 2025-02-11 12:20:39 -06:00 committed by GitHub
parent 46d19d82c2
commit 33f9d66e81
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 32 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

38
esphome/components/voice_assistant/voice_assistant.cpp
View File

@ -1,4 +1,5 @@
#include "voice_assistant.h"
#include "esphome/core/defines.h"
#ifdef USE_VOICE_ASSISTANT
@ -127,7 +128,7 @@ void VoiceAssistant::clear_buffers_() {
}
#ifdef USE_SPEAKER
if (this->speaker_buffer_ != nullptr) {
if ((this->speaker_ != nullptr) && (this->speaker_buffer_ != nullptr)) {
memset(this->speaker_buffer_, 0, SPEAKER_BUFFER_SIZE);
this->speaker_buffer_size_ = 0;
@ -159,7 +160,7 @@ void VoiceAssistant::deallocate_buffers_() {
this->input_buffer_ = nullptr;
#ifdef USE_SPEAKER
if (this->speaker_buffer_ != nullptr) {
if ((this->speaker_ != nullptr) && (this->speaker_buffer_ != nullptr)) {
ExternalRAMAllocator<uint8_t> speaker_deallocator(ExternalRAMAllocator<uint8_t>::ALLOW_FAILURE);
speaker_deallocator.deallocate(this->speaker_buffer_, SPEAKER_BUFFER_SIZE);
this->speaker_buffer_ = nullptr;
@ -389,14 +390,7 @@ void VoiceAssistant::loop() {
}
#endif
if (playing) {
this->set_timeout("playing", 2000, [this]() {
this->cancel_timeout("speaker-timeout");
this->set_state_(State::IDLE, State::IDLE);
api::VoiceAssistantAnnounceFinished msg;
msg.success = true;
this->api_client_->send_voice_assistant_announce_finished(msg);
});
this->start_playback_timeout_();
}
break;
}
@ -614,6 +608,8 @@ void VoiceAssistant::request_stop() {
this->desired_state_ = State::IDLE;
break;
case State::AWAITING_RESPONSE:
this->signal_stop_();
break;
case State::STREAMING_RESPONSE:
case State::RESPONSE_FINISHED:
break; // Let the incoming audio stream finish then it will go to idle.
@ -631,6 +627,17 @@ void VoiceAssistant::signal_stop_() {
this->api_client_->send_voice_assistant_request(msg);
}
void VoiceAssistant::start_playback_timeout_() {
this->set_timeout("playing", 100, [this]() {
this->cancel_timeout("speaker-timeout");
this->set_state_(State::IDLE, State::IDLE);
api::VoiceAssistantAnnounceFinished msg;
msg.success = true;
this->api_client_->send_voice_assistant_announce_finished(msg);
});
}
void VoiceAssistant::on_event(const api::VoiceAssistantEventResponse &msg) {
ESP_LOGD(TAG, "Event Type: %" PRId32, msg.event_type);
switch (msg.event_type) {
@ -715,6 +722,8 @@ void VoiceAssistant::on_event(const api::VoiceAssistantEventResponse &msg) {
#ifdef USE_MEDIA_PLAYER
if (this->media_player_ != nullptr) {
this->media_player_->make_call().set_media_url(url).set_announcement(true).perform();
// Start the playback timeout, as the media player state isn't immediately updated
this->start_playback_timeout_();
}
#endif
this->tts_end_trigger_->trigger(url);
@ -725,7 +734,11 @@ void VoiceAssistant::on_event(const api::VoiceAssistantEventResponse &msg) {
}
case api::enums::VOICE_ASSISTANT_RUN_END: {
ESP_LOGD(TAG, "Assist Pipeline ended");
if (this->state_ == State::STREAMING_MICROPHONE) {
if ((this->state_ == State::STARTING_PIPELINE) || (this->state_ == State::AWAITING_RESPONSE)) {
// Pipeline ended before starting microphone
// Or there wasn't a TTS start event ("nevermind")
this->set_state_(State::IDLE, State::IDLE);
} else if (this->state_ == State::STREAMING_MICROPHONE) {
this->ring_buffer_->reset();
#ifdef USE_ESP_ADF
if (this->use_wake_word_) {
@ -736,9 +749,6 @@ void VoiceAssistant::on_event(const api::VoiceAssistantEventResponse &msg) {
{
this->set_state_(State::IDLE, State::IDLE);
}
} else if (this->state_ == State::AWAITING_RESPONSE) {
// No TTS start event ("nevermind")
this->set_state_(State::IDLE, State::IDLE);
}
this->defer([this]() { this->end_trigger_->trigger(); });
break;

8
esphome/components/voice_assistant/voice_assistant.h
View File

@ -40,6 +40,7 @@ enum VoiceAssistantFeature : uint32_t {
FEATURE_SPEAKER = 1 << 1,
FEATURE_API_AUDIO = 1 << 2,
FEATURE_TIMERS = 1 << 3,
FEATURE_ANNOUNCE = 1 << 4,
};
enum class State {
@ -136,6 +137,12 @@ class VoiceAssistant : public Component {
flags |= VoiceAssistantFeature::FEATURE_TIMERS;
}
#ifdef USE_MEDIA_PLAYER
if (this->media_player_ != nullptr) {
flags |= VoiceAssistantFeature::FEATURE_ANNOUNCE;
}
#endif
return flags;
}
@ -209,6 +216,7 @@ class VoiceAssistant : public Component {
void set_state_(State state);
void set_state_(State state, State desired_state);
void signal_stop_();
void start_playback_timeout_();
std::unique_ptr<socket::Socket> socket_ = nullptr;
struct sockaddr_storage dest_addr_;