sirmrak/linux-voice-assistant
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge pull request #38 from imonlinux/media_player_persistence

Browse Source 
Fix ESPHome media_player persistence and entity key stability
This commit is contained in:
imonlinux
2025-11-20 13:05:45 -08:00
committed by GitHub
parent f34bc90edd 0b3c3e7b42
commit 8d928c6b02
1 changed files with 105 additions and 59 deletions
Download Patch File Download Diff File Expand all files Collapse all files
linux_voice_assistant/satellite.py
+105 -59
View File
@@ -53,18 +53,52 @@ class VoiceSatelliteProtocol(APIServer):
def __init__(self, state: ServerState) -> None: def __init__(self, state: ServerState) -> None:
super().__init__(state.name) super().__init__(state.name)
self.state = state self.state = state
self.state.satellite = self self.state.satellite = self
self.media_player_entity = MediaPlayerEntity( # --- Ensure exactly one MediaPlayerEntity with a stable key ---
server=self, if self.state.entities:
state=state, existing = self.state.entities[0]
key=len(state.entities), if not isinstance(existing, MediaPlayerEntity):
name="Media Player", _LOGGER.warning(
object_id="linux_voice_assistant_media_player", "First ESPHome entity is not MediaPlayerEntity (%r). "
music_player=state.music_player, "Replacing it with a new MediaPlayerEntity.",
announce_player=state.tts_player, type(existing),
) )
self.state.entities.append(self.media_player_entity) self.media_player_entity = MediaPlayerEntity(
server=self,
state=state,
key=0,
name="Media Player",
object_id="linux_voice_assistant_media_player",
music_player=state.music_player,
announce_player=state.tts_player,
)
self.state.entities[0] = self.media_player_entity
else:
# Reuse existing entity but bind it to the current protocol
self.media_player_entity = existing
self.media_player_entity.server = self
self.media_player_entity.key = 0
else:
# First connection in this process: create the media player once
self.media_player_entity = MediaPlayerEntity(
server=self,
state=state,
key=0,
name="Media Player",
object_id="linux_voice_assistant_media_player",
music_player=state.music_player,
announce_player=state.tts_player,
)
self.state.entities.append(self.media_player_entity)
# If more entities somehow accumulated, prune them to avoid confusing HA
if len(self.state.entities) > 1:
_LOGGER.warning(
"Pruning %d extra ESPHome entities; keeping only the first.",
len(self.state.entities) - 1,
)
del self.state.entities[1:]
# State machine # State machine
self._state: SatelliteState = SatelliteState.STARTING self._state: SatelliteState = SatelliteState.STARTING
@@ -76,15 +110,16 @@ class VoiceSatelliteProtocol(APIServer):
self._run_end_received: bool = False self._run_end_received: bool = False
self._tts_end_received: bool = False self._tts_end_received: bool = False
self._is_announcement: bool = False # <--- FIX: Track if current audio is an announcement # Track if current audio is an announcement
self._is_announcement: bool = False
self._external_wake_words: Dict[str, VoiceAssistantExternalWakeWord] = {} self._external_wake_words: Dict[str, VoiceAssistantExternalWakeWord] = {}
def _set_state(self, new_state: SatelliteState): def _set_state(self, new_state: SatelliteState):
if self._state == new_state: if self._state == new_state:
return return
_LOGGER.debug(f"State transition: {self._state} -> {new_state}") _LOGGER.debug("State transition: %s -> %s", self._state, new_state)
self._state = new_state self._state = new_state
if new_state == SatelliteState.IDLE: if new_state == SatelliteState.IDLE:
@@ -110,21 +145,22 @@ class VoiceSatelliteProtocol(APIServer):
if event_type == VoiceAssistantEventType.VOICE_ASSISTANT_RUN_START: if event_type == VoiceAssistantEventType.VOICE_ASSISTANT_RUN_START:
self._run_end_received = False self._run_end_received = False
self._tts_end_received = False self._tts_end_received = False
self._is_announcement = False # <--- Reset announcement flag on new run self._is_announcement = False
self._tts_url = data.get("url") self._tts_url = data.get("url")
self._continue_conversation = False self._continue_conversation = False
elif event_type == VoiceAssistantEventType.VOICE_ASSISTANT_STT_START: elif event_type == VoiceAssistantEventType.VOICE_ASSISTANT_STT_START:
self._set_state(SatelliteState.LISTENING) self._set_state(SatelliteState.LISTENING)
elif event_type == VoiceAssistantEventType.VOICE_ASSISTANT_STT_VAD_START: elif event_type == VoiceAssistantEventType.VOICE_ASSISTANT_STT_VAD_START:
self.state.event_bus.publish("voice_vad_start", data) self.state.event_bus.publish("voice_vad_start", data)
elif event_type == VoiceAssistantEventType.VOICE_ASSISTANT_STT_END: elif event_type == VoiceAssistantEventType.VOICE_ASSISTANT_STT_END:
self._is_streaming_audio = False self._is_streaming_audio = False
self._set_state(SatelliteState.THINKING) self._set_state(SatelliteState.THINKING)
elif event_type == VoiceAssistantEventType.VOICE_ASSISTANT_STT_VAD_END: elif event_type == VoiceAssistantEventType.VOICE_ASSISTANT_STT_VAD_END:
# No-op for now; could be used for LED cues
pass pass
elif event_type == VoiceAssistantEventType.VOICE_ASSISTANT_INTENT_END: elif event_type == VoiceAssistantEventType.VOICE_ASSISTANT_INTENT_END:
@@ -138,7 +174,7 @@ class VoiceSatelliteProtocol(APIServer):
self._tts_url = data.get("url") self._tts_url = data.get("url")
self._tts_end_received = True self._tts_end_received = True
self.play_tts() self.play_tts()
elif event_type == VoiceAssistantEventType.VOICE_ASSISTANT_RUN_END: elif event_type == VoiceAssistantEventType.VOICE_ASSISTANT_RUN_END:
self._run_end_received = True self._run_end_received = True
if self._state != SatelliteState.RESPONDING: if self._state != SatelliteState.RESPONDING:
@@ -148,7 +184,6 @@ class VoiceSatelliteProtocol(APIServer):
self._set_state(SatelliteState.ERROR) self._set_state(SatelliteState.ERROR)
self.state.loop.call_later(5.0, self._set_state, SatelliteState.IDLE) self.state.loop.call_later(5.0, self._set_state, SatelliteState.IDLE)
def handle_timer_event( def handle_timer_event(
self, self,
event_type: VoiceAssistantTimerEventType, event_type: VoiceAssistantTimerEventType,
@@ -172,15 +207,15 @@ class VoiceSatelliteProtocol(APIServer):
for arg in msg.data: for arg in msg.data:
data[arg.name] = arg.value data[arg.name] = arg.value
self.handle_voice_event(VoiceAssistantEventType(msg.event_type), data) self.handle_voice_event(VoiceAssistantEventType(msg.event_type), data)
elif isinstance(msg, VoiceAssistantAnnounceRequest): elif isinstance(msg, VoiceAssistantAnnounceRequest):
_LOGGER.debug("Announcing: %s", msg.text) _LOGGER.debug("Announcing: %s", msg.text)
urls = [] urls: List[str] = []
if msg.preannounce_media_id: if msg.preannounce_media_id:
urls.append(msg.preannounce_media_id) urls.append(msg.preannounce_media_id)
urls.append(msg.media_id) urls.append(msg.media_id)
self._is_announcement = True # <--- FIX: Mark this as an announcement self._is_announcement = True
self.state.active_wake_words.add(self.state.stop_word.id) self.state.active_wake_words.add(self.state.stop_word.id)
self._continue_conversation = msg.start_conversation self._continue_conversation = msg.start_conversation
self.duck() self.duck()
@@ -188,10 +223,10 @@ class VoiceSatelliteProtocol(APIServer):
yield from self.media_player_entity.play( yield from self.media_player_entity.play(
urls, announcement=True, done_callback=self._tts_finished urls, announcement=True, done_callback=self._tts_finished
) )
elif isinstance(msg, VoiceAssistantTimerEventResponse): elif isinstance(msg, VoiceAssistantTimerEventResponse):
self.handle_timer_event(VoiceAssistantTimerEventType(msg.event_type), msg) self.handle_timer_event(VoiceAssistantTimerEventType(msg.event_type), msg)
elif isinstance(msg, DeviceInfoRequest): elif isinstance(msg, DeviceInfoRequest):
yield DeviceInfoResponse( yield DeviceInfoResponse(
uses_password=False, uses_password=False,
@@ -205,7 +240,7 @@ class VoiceSatelliteProtocol(APIServer):
| VoiceAssistantFeature.TIMERS | VoiceAssistantFeature.TIMERS
), ),
) )
elif isinstance( elif isinstance(
msg, msg,
( (
@@ -219,9 +254,9 @@ class VoiceSatelliteProtocol(APIServer):
if isinstance(msg, ListEntitiesRequest): if isinstance(msg, ListEntitiesRequest):
yield ListEntitiesDoneResponse() yield ListEntitiesDoneResponse()
elif isinstance(msg, VoiceAssistantConfigurationRequest): elif isinstance(msg, VoiceAssistantConfigurationRequest):
available_wake_words = [ available_wake_words: List[VoiceAssistantWakeWord] = [
VoiceAssistantWakeWord( VoiceAssistantWakeWord(
id=ww.id, id=ww.id,
wake_word=ww.wake_word, wake_word=ww.wake_word,
@@ -230,10 +265,15 @@ class VoiceSatelliteProtocol(APIServer):
for ww in self.state.available_wake_words.values() for ww in self.state.available_wake_words.values()
] ]
# Reset external wake words cache and add new ones
self._external_wake_words.clear() self._external_wake_words.clear()
for eww in msg.external_wake_words: for eww in msg.external_wake_words:
if eww.model_type != "micro": if eww.model_type != "micro":
_LOGGER.warning("Skipping external wake word %s (type=%s)", eww.id, eww.model_type) _LOGGER.warning(
"Skipping external wake word %s (type=%s)",
eww.id,
eww.model_type,
)
continue continue
available_wake_words.append( available_wake_words.append(
@@ -244,36 +284,39 @@ class VoiceSatelliteProtocol(APIServer):
) )
) )
self._external_wake_words[eww.id] = eww self._external_wake_words[eww.id] = eww
yield VoiceAssistantConfigurationResponse( yield VoiceAssistantConfigurationResponse(
available_wake_words=available_wake_words, available_wake_words=available_wake_words,
active_wake_words=[ active_wake_words=[
ww.id for ww in self.state.wake_words.values() ww.id
for ww in self.state.wake_words.values()
if ww.id in self.state.active_wake_words if ww.id in self.state.active_wake_words
], ],
max_active_wake_words=2, max_active_wake_words=2,
) )
_LOGGER.info("Connected to Home Assistant") _LOGGER.info("Connected to Home Assistant")
self._set_state(SatelliteState.IDLE) self._set_state(SatelliteState.IDLE)
self.state.event_bus.publish("ha_connected") self.state.event_bus.publish("ha_connected")
elif isinstance(msg, VoiceAssistantSetConfiguration): elif isinstance(msg, VoiceAssistantSetConfiguration):
# FIX: Offload to background task to prevent blocking the loop during download # Offload to background task to prevent blocking the loop during download
self.state.loop.create_task(self._handle_set_configuration_task(msg)) self.state.loop.create_task(self._handle_set_configuration_task(msg))
# Yield nothing immediately; response will be sent asynchronously # Yield nothing immediately; response will be sent asynchronously
async def _handle_set_configuration_task(self, msg: VoiceAssistantSetConfiguration) -> None: async def _handle_set_configuration_task(
self, msg: VoiceAssistantSetConfiguration
) -> None:
"""Asynchronous handler for SetConfiguration to avoid blocking I/O.""" """Asynchronous handler for SetConfiguration to avoid blocking I/O."""
active_wake_words: Set[str] = set() active_wake_words: Set[str] = set()
for wake_word_id in msg.active_wake_words: for wake_word_id in msg.active_wake_words:
if wake_word_id in self.state.wake_words: if wake_word_id in self.state.wake_words:
active_wake_words.add(wake_word_id) active_wake_words.add(wake_word_id)
continue continue
model_info = self.state.available_wake_words.get(wake_word_id) model_info = self.state.available_wake_words.get(wake_word_id)
if not model_info: if not model_info:
external_wake_word = self._external_wake_words.get(wake_word_id) external_wake_word = self._external_wake_words.get(wake_word_id)
if not external_wake_word: if not external_wake_word:
@@ -289,14 +332,14 @@ class VoiceSatelliteProtocol(APIServer):
_LOGGER.debug("Loading wake word: %s", model_info.wake_word_path) _LOGGER.debug("Loading wake word: %s", model_info.wake_word_path)
self.state.wake_words[wake_word_id] = model_info.load() self.state.wake_words[wake_word_id] = model_info.load()
_LOGGER.info("Wake word set: %s", wake_word_id) _LOGGER.info("Wake word set: %s", wake_word_id)
active_wake_words.add(wake_word_id) active_wake_words.add(wake_word_id)
break break
self.state.active_wake_words = active_wake_words self.state.active_wake_words = active_wake_words
_LOGGER.debug("Active wake words: %s", active_wake_words) _LOGGER.debug("Active wake words: %s", active_wake_words)
self.state.preferences.active_wake_words = list(active_wake_words) self.state.preferences.active_wake_words = list(active_wake_words)
self.state.save_preferences() self.state.save_preferences()
self.state.wake_words_changed = True self.state.wake_words_changed = True
@@ -308,13 +351,13 @@ class VoiceSatelliteProtocol(APIServer):
def wakeup(self, wake_word: Union[MicroWakeWord, OpenWakeWord]) -> None: def wakeup(self, wake_word: Union[MicroWakeWord, OpenWakeWord]) -> None:
if self._state not in (SatelliteState.IDLE, SatelliteState.STARTING): if self._state not in (SatelliteState.IDLE, SatelliteState.STARTING):
return return
if self._timer_finished: if self._timer_finished:
self._timer_finished = False self._timer_finished = False
self.state.tts_player.stop() self.state.tts_player.stop()
_LOGGER.debug("Stopping timer finished sound") _LOGGER.debug("Stopping timer finished sound")
return return
wake_word_phrase = getattr(wake_word, "wake_word", "wake word") wake_word_phrase = getattr(wake_word, "wake_word", "wake word")
_LOGGER.debug("Detected wake word: %s", wake_word_phrase) _LOGGER.debug("Detected wake word: %s", wake_word_phrase)
self.send_messages( self.send_messages(
@@ -359,7 +402,7 @@ class VoiceSatelliteProtocol(APIServer):
self._set_state(SatelliteState.LISTENING) self._set_state(SatelliteState.LISTENING)
else: else:
self._set_state(SatelliteState.IDLE) self._set_state(SatelliteState.IDLE)
_LOGGER.debug("Final state determined") _LOGGER.debug("Final state determined")
def _tts_finished(self) -> None: def _tts_finished(self) -> None:
@@ -367,8 +410,8 @@ class VoiceSatelliteProtocol(APIServer):
self.send_messages([VoiceAssistantAnnounceFinished()]) self.send_messages([VoiceAssistantAnnounceFinished()])
_LOGGER.debug("TTS audio playback finished") _LOGGER.debug("TTS audio playback finished")
# FIX: If this was just an announcement, we are done. # If this was just an announcement, we are done,
# Or if we received the official Run End. # or if we received the official Run End.
if self._is_announcement or self._run_end_received: if self._is_announcement or self._run_end_received:
self._determine_final_state() self._determine_final_state()
self._is_announcement = False self._is_announcement = False
@@ -376,7 +419,6 @@ class VoiceSatelliteProtocol(APIServer):
if self._state == SatelliteState.RESPONDING: if self._state == SatelliteState.RESPONDING:
self._set_state(SatelliteState.THINKING) self._set_state(SatelliteState.THINKING)
def _play_timer_finished(self) -> None: def _play_timer_finished(self) -> None:
if not self._timer_finished: if not self._timer_finished:
self.unduck() self.unduck()
@@ -393,8 +435,8 @@ class VoiceSatelliteProtocol(APIServer):
) -> Optional[AvailableWakeWord]: ) -> Optional[AvailableWakeWord]:
"""Wrapper to run the blocking download in a thread executor.""" """Wrapper to run the blocking download in a thread executor."""
return await self.state.loop.run_in_executor( return await self.state.loop.run_in_executor(
None, None,
functools.partial(self._download_external_wake_word_sync, external_wake_word) functools.partial(self._download_external_wake_word_sync, external_wake_word),
) )
def _download_external_wake_word_sync( def _download_external_wake_word_sync(
@@ -423,7 +465,9 @@ class VoiceSatelliteProtocol(APIServer):
) )
if should_download_config or should_download_model: if should_download_config or should_download_model:
_LOGGER.debug("Downloading %s to %s", external_wake_word.url, config_path) _LOGGER.debug(
"Downloading %s to %s", external_wake_word.url, config_path
)
try: try:
with urlopen(external_wake_word.url, timeout=10) as request: with urlopen(external_wake_word.url, timeout=10) as request:
if request.status != 200: if request.status != 200:
@@ -436,9 +480,9 @@ class VoiceSatelliteProtocol(APIServer):
with open(config_path, "wb") as model_file: with open(config_path, "wb") as model_file:
shutil.copyfileobj(request, model_file) shutil.copyfileobj(request, model_file)
except Exception as e: except Exception as exc: # noqa: BLE001
_LOGGER.error("Exception downloading config: %s", e) _LOGGER.error("Exception downloading config: %s", exc)
return None return None
if should_download_model: if should_download_model:
parsed_url = urlparse(external_wake_word.url) parsed_url = urlparse(external_wake_word.url)
@@ -452,15 +496,17 @@ class VoiceSatelliteProtocol(APIServer):
with urlopen(model_url, timeout=10) as request: with urlopen(model_url, timeout=10) as request:
if request.status != 200: if request.status != 200:
_LOGGER.warning( _LOGGER.warning(
"Failed to download: %s, status=%s", model_url, request.status "Failed to download: %s, status=%s",
model_url,
request.status,
) )
return None return None
with open(model_path, "wb") as model_file: with open(model_path, "wb") as model_file:
shutil.copyfileobj(request, model_file) shutil.copyfileobj(request, model_file)
except Exception as e: except Exception as exc: # noqa: BLE001
_LOGGER.error("Exception downloading model: %s", e) _LOGGER.error("Exception downloading model: %s", exc)
return None return None
return AvailableWakeWord( return AvailableWakeWord(
id=external_wake_word.id, id=external_wake_word.id,