Merge branch 'main' into main

linux_voice_assistant/__main__.py

@@ -63,6 +63,11 @@ async def main() -> None:
         "--wake-model", default="okay_nabu", help="Id of active wake model"
     )
     parser.add_argument("--stop-model", default="stop", help="Id of stop model")
+    parser.add_argument(
+        "--download-dir",
+        default=_REPO_DIR / "local",
+        help="Directory to download custom wake word models, etc.",
+    )
     parser.add_argument(
         "--refractory-seconds",
         default=2.0,
@@ -120,6 +125,9 @@ async def main() -> None:
     logging.basicConfig(level=logging.DEBUG if args.debug else logging.INFO)
     _LOGGER.debug(args)
 
+    args.download_dir = Path(args.download_dir)
+    args.download_dir.mkdir(parents=True, exist_ok=True)
+
     # Resolve microphone
     if args.audio_input_device is not None:
         try:
@@ -133,6 +141,7 @@ async def main() -> None:
 
     # Load available wake words
     wake_word_dirs = [Path(ww_dir) for ww_dir in args.wake_word_dir]
+    wake_word_dirs.append(args.download_dir / "external_wake_words")
     available_wake_words: Dict[str, AvailableWakeWord] = {}
 
     for wake_word_dir in wake_word_dirs:
@@ -227,6 +236,7 @@ async def main() -> None:
         preferences=preferences,
         preferences_path=preferences_path,
         refractory_seconds=args.refractory_seconds,
+        download_dir=args.download_dir,
     )
 
     if args.enable_thinking_sound:

linux_voice_assistant/api_server.py

@@ -9,8 +9,8 @@ from typing import TYPE_CHECKING, List, Optional
 # pylint: disable=no-name-in-module
 from aioesphomeapi._frame_helper.packets import make_plain_text_packets
 from aioesphomeapi.api_pb2 import (  # type: ignore[attr-defined]
-    ConnectRequest,
-    ConnectResponse,
+    AuthenticationRequest,
+    AuthenticationResponse,
     DisconnectRequest,
     DisconnectResponse,
     HelloRequest,
@@ -57,8 +57,8 @@ class APIServer(asyncio.Protocol):
             )
             return
 
-        if isinstance(msg_inst, ConnectRequest):
-            self.send_messages([ConnectResponse()])
+        if isinstance(msg_inst, AuthenticationRequest):
+            self.send_messages([AuthenticationResponse()])
         elif isinstance(msg_inst, DisconnectRequest):
             self.send_messages([DisconnectResponse()])
             _LOGGER.debug("Disconnect requested")

linux_voice_assistant/models.py

@@ -72,6 +72,7 @@ class ServerState:
     timer_finished_sound: str
     preferences: Preferences
     preferences_path: Path
+    download_dir: Path
 
     media_player_entity: "Optional[MediaPlayerEntity]" = None
     satellite: "Optional[VoiceSatelliteProtocol]" = None

linux_voice_assistant/satellite.py

@@ -1,9 +1,14 @@
 """Voice satellite protocol."""
 
+import hashlib
 import logging
+import posixpath
+import shutil
 import time
 from collections.abc import Iterable
 from typing import Dict, Optional, Set, Union
+from urllib.parse import urlparse, urlunparse
+from urllib.request import urlopen
 
 # pylint: disable=no-name-in-module
 from aioesphomeapi.api_pb2 import (  # type: ignore[attr-defined]
@@ -19,6 +24,7 @@ from aioesphomeapi.api_pb2 import (  # type: ignore[attr-defined]
     VoiceAssistantConfigurationRequest,
     VoiceAssistantConfigurationResponse,
     VoiceAssistantEventResponse,
+    VoiceAssistantExternalWakeWord,
     VoiceAssistantRequest,
     VoiceAssistantSetConfiguration,
     VoiceAssistantTimerEventResponse,
@@ -34,9 +40,8 @@ from pymicro_wakeword import MicroWakeWord
 from pyopen_wakeword import OpenWakeWord
 
 from .api_server import APIServer
-from .entity import MediaPlayerEntity
 from .entity import MediaPlayerEntity, ThinkingSoundEntity
-from .models import ServerState
+from .models import AvailableWakeWord, ServerState, WakeWordType
 from .util import call_all
 
 _LOGGER = logging.getLogger(__name__)
@@ -104,6 +109,7 @@ class VoiceSatelliteProtocol(APIServer):
         self._continue_conversation = False
         self._timer_finished = False
         self._processing = False
+        self._external_wake_words: Dict[str, VoiceAssistantExternalWakeWord] = {}
 
     def _set_thinking_sound_enabled(self, new_state: bool) -> None:
         self.state.thinking_sound_enabled = bool(new_state)
@@ -115,6 +121,7 @@ class VoiceSatelliteProtocol(APIServer):
             _LOGGER.debug("Thinking sound disabled")
             pass
         self.state.save_preferences()
+        
 
     def handle_voice_event(
         self, event_type: VoiceAssistantEventType, data: Dict[str, str]
@@ -226,15 +233,31 @@ class VoiceSatelliteProtocol(APIServer):
             if isinstance(msg, ListEntitiesRequest):
                 yield ListEntitiesDoneResponse()
         elif isinstance(msg, VoiceAssistantConfigurationRequest):
-            yield VoiceAssistantConfigurationResponse(
-                available_wake_words=[
+            available_wake_words = [
+                VoiceAssistantWakeWord(
+                    id=ww.id,
+                    wake_word=ww.wake_word,
+                    trained_languages=ww.trained_languages,
+                )
+                for ww in self.state.available_wake_words.values()
+            ]
+
+            for eww in msg.external_wake_words:
+                if eww.model_type != "micro":
+                    continue
+
+                available_wake_words.append(
                     VoiceAssistantWakeWord(
-                        id=ww.id,
-                        wake_word=ww.wake_word,
-                        trained_languages=ww.trained_languages,
+                        id=eww.id,
+                        wake_word=eww.wake_word,
+                        trained_languages=eww.trained_languages,
                     )
-                    for ww in self.state.available_wake_words.values()
-                ],
+                )
+
+                self._external_wake_words[eww.id] = eww
+
+            yield VoiceAssistantConfigurationResponse(
+                available_wake_words=available_wake_words,
                 active_wake_words=[
                     ww.id
                     for ww in self.state.wake_words.values()
@@ -255,7 +278,16 @@ class VoiceSatelliteProtocol(APIServer):
 
                 model_info = self.state.available_wake_words.get(wake_word_id)
                 if not model_info:
-                    continue
+                    # Check external wake words (may require download)
+                    external_wake_word = self._external_wake_words.get(wake_word_id)
+                    if not external_wake_word:
+                        continue
+
+                    model_info = self._download_external_wake_word(external_wake_word)
+                    if not model_info:
+                        continue
+
+                    self.state.available_wake_words[wake_word_id] = model_info
 
                 _LOGGER.debug("Loading wake word: %s", model_info.wake_word_path)
                 self.state.wake_words[wake_word_id] = model_info.load()
@@ -352,3 +384,70 @@ class VoiceSatelliteProtocol(APIServer):
     def connection_lost(self, exc):
         super().connection_lost(exc)
         _LOGGER.info("Disconnected from Home Assistant")
+
+    def _download_external_wake_word(
+        self, external_wake_word: VoiceAssistantExternalWakeWord
+    ) -> Optional[AvailableWakeWord]:
+        eww_dir = self.state.download_dir / "external_wake_words"
+        eww_dir.mkdir(parents=True, exist_ok=True)
+
+        config_path = eww_dir / f"{external_wake_word.id}.json"
+        should_download_config = not config_path.exists()
+
+        # Check if we need to download the model file
+        model_path = eww_dir / f"{external_wake_word.id}.tflite"
+        should_download_model = True
+        if model_path.exists():
+            model_size = model_path.stat().st_size
+            if model_size == external_wake_word.model_size:
+                with open(model_path, "rb") as model_file:
+                    model_hash = hashlib.sha256(model_file.read()).hexdigest()
+
+                if model_hash == external_wake_word.model_hash:
+                    should_download_model = False
+                    _LOGGER.debug(
+                        "Model size and hash match for %s. Skipping download.",
+                        external_wake_word.id,
+                    )
+
+        if should_download_config or should_download_model:
+            # Download config
+            _LOGGER.debug("Downloading %s to %s", external_wake_word.url, config_path)
+            with urlopen(external_wake_word.url) as request:
+                if request.status != 200:
+                    _LOGGER.warning(
+                        "Failed to download: %s, status=%s",
+                        external_wake_word.url,
+                        request.status,
+                    )
+                    return None
+
+                with open(config_path, "wb") as model_file:
+                    shutil.copyfileobj(request, model_file)
+
+        if should_download_model:
+            # Download model file
+            parsed_url = urlparse(external_wake_word.url)
+            parsed_url = parsed_url._replace(
+                path=posixpath.join(posixpath.dirname(parsed_url.path), model_path.name)
+            )
+            model_url = urlunparse(parsed_url)
+
+            _LOGGER.debug("Downloading %s to %s", model_url, model_path)
+            with urlopen(model_url) as request:
+                if request.status != 200:
+                    _LOGGER.warning(
+                        "Failed to download: %s, status=%s", model_url, request.status
+                    )
+                    return None
+
+                with open(model_path, "wb") as model_file:
+                    shutil.copyfileobj(request, model_file)
+
+        return AvailableWakeWord(
+            id=external_wake_word.id,
+            type=WakeWordType.MICRO_WAKE_WORD,
+            wake_word=external_wake_word.wake_word,
+            trained_languages=external_wake_word.trained_languages,
+            wake_word_path=config_path,
+        )