livekit-plugins-google 0.9.0__py3-none-any.whl → 0.9.1__py3-none-any.whl

livekit/plugins/google/beta/realtime/realtime_api.py

@@ -105,7 +105,7 @@ class RealtimeModel:
 
         Args:
             instructions (str, optional): Initial system instructions for the model. Defaults to "".
-            api_key (str or None, optional): OpenAI API key. If None, will attempt to read from the environment variable OPENAI_API_KEY
+            api_key (str or None, optional): Google Gemini API key. If None, will attempt to read from the environment variable GOOGLE_API_KEY.
             modalities (ResponseModality): Modalities to use, such as ["TEXT", "AUDIO"]. Defaults to ["AUDIO"].
             model (str or None, optional): The name of the model to use. Defaults to "gemini-2.0-flash-exp".
             voice (api_proto.Voice, optional): Voice setting for audio outputs. Defaults to "Puck".

livekit/plugins/google/stt.py

@@ -16,6 +16,7 @@ from __future__ import annotations
 
 import asyncio
 import dataclasses
+import time
 import weakref
 from dataclasses import dataclass
 from typing import List, Union
@@ -44,6 +45,10 @@ from .models import SpeechLanguages, SpeechModels
 LgType = Union[SpeechLanguages, str]
 LanguageCode = Union[LgType, List[LgType]]
 
+# Google STT has a timeout of 5 mins, we'll attempt to restart the session
+# before that timeout is reached
+_max_session_duration = 240
+
 
 # This class is only be used internally to encapsulate the options
 @dataclass
@@ -229,8 +234,6 @@ class STT(stt.STT):
             raise APIStatusError(
                 e.message,
                 status_code=e.code or -1,
-                request_id=None,
-                body=None,
             )
         except Exception as e:
             raise APIConnectionError() from e
@@ -278,6 +281,13 @@ class STT(stt.STT):
             self._config.spoken_punctuation = spoken_punctuation
         if model is not None:
             self._config.model = model
+        client = None
+        recognizer = None
+        if location is not None:
+            self._location = location
+            # if location is changed, fetch a new client and recognizer as per the new location
+            client = self._ensure_client()
+            recognizer = self._recognizer
         if keywords is not None:
             self._config.keywords = keywords
 
@@ -289,8 +299,9 @@ class STT(stt.STT):
                 punctuate=punctuate,
                 spoken_punctuation=spoken_punctuation,
                 model=model,
-                location=location,
                 keywords=keywords,
+                client=client,
+                recognizer=recognizer,
             )
 
 
@@ -312,6 +323,7 @@ class SpeechStream(stt.SpeechStream):
         self._recognizer = recognizer
         self._config = config
         self._reconnect_event = asyncio.Event()
+        self._session_connected_at: float = 0
 
     def update_options(
         self,
@@ -322,8 +334,9 @@ class SpeechStream(stt.SpeechStream):
         punctuate: bool | None = None,
         spoken_punctuation: bool | None = None,
         model: SpeechModels | None = None,
-        location: str | None = None,
         keywords: List[tuple[str, float]] | None = None,
+        client: SpeechAsyncClient | None = None,
+        recognizer: str | None = None,
     ):
         if languages is not None:
             if isinstance(languages, str):
@@ -341,13 +354,17 @@ class SpeechStream(stt.SpeechStream):
             self._config.model = model
         if keywords is not None:
             self._config.keywords = keywords
+        if client is not None:
+            self._client = client
+        if recognizer is not None:
+            self._recognizer = recognizer
 
         self._reconnect_event.set()
 
     async def _run(self) -> None:
         # google requires a async generator when calling streaming_recognize
         # this function basically convert the queue into a async generator
-        async def input_generator():
+        async def input_generator(should_stop: asyncio.Event):
             try:
                 # first request should contain the config
                 yield cloud_speech.StreamingRecognizeRequest(
@@ -356,6 +373,12 @@ class SpeechStream(stt.SpeechStream):
                 )
 
                 async for frame in self._input_ch:
+                    # when the stream is aborted due to reconnect, this input_generator
+                    # needs to stop consuming frames
+                    # when the generator stops, the previous gRPC stream will close
+                    if should_stop.is_set():
+                        return
+
                     if isinstance(frame, rtc.AudioFrame):
                         yield cloud_speech.StreamingRecognizeRequest(
                             audio=frame.data.tobytes()
@@ -367,6 +390,7 @@ class SpeechStream(stt.SpeechStream):
                 )
 
         async def process_stream(stream):
+            has_started = False
             async for resp in stream:
                 if (
                     resp.speech_event_type
@@ -375,6 +399,7 @@ class SpeechStream(stt.SpeechStream):
                     self._event_ch.send_nowait(
                         stt.SpeechEvent(type=stt.SpeechEventType.START_OF_SPEECH)
                     )
+                    has_started = True
 
                 if (
                     resp.speech_event_type
@@ -399,6 +424,22 @@ class SpeechStream(stt.SpeechStream):
                                 alternatives=[speech_data],
                             )
                         )
+                        if (
+                            time.time() - self._session_connected_at
+                            > _max_session_duration
+                        ):
+                            logger.debug(
+                                "Google STT maximum connection time reached. Reconnecting..."
+                            )
+                            if has_started:
+                                self._event_ch.send_nowait(
+                                    stt.SpeechEvent(
+                                        type=stt.SpeechEventType.END_OF_SPEECH
+                                    )
+                                )
+                                has_started = False
+                            self._reconnect_event.set()
+                            return
 
                 if (
                     resp.speech_event_type
@@ -407,6 +448,7 @@ class SpeechStream(stt.SpeechStream):
                     self._event_ch.send_nowait(
                         stt.SpeechEvent(type=stt.SpeechEventType.END_OF_SPEECH)
                     )
+                    has_started = False
 
         while True:
             try:
@@ -431,12 +473,15 @@ class SpeechStream(stt.SpeechStream):
                     ),
                 )
 
+                should_stop = asyncio.Event()
                 stream = await self._client.streaming_recognize(
-                    requests=input_generator(),
+                    requests=input_generator(should_stop),
                 )
+                self._session_connected_at = time.time()
 
                 process_stream_task = asyncio.create_task(process_stream(stream))
                 wait_reconnect_task = asyncio.create_task(self._reconnect_event.wait())
+
                 try:
                     done, _ = await asyncio.wait(
                         [process_stream_task, wait_reconnect_task],
@@ -445,14 +490,23 @@ class SpeechStream(stt.SpeechStream):
                     for task in done:
                         if task != wait_reconnect_task:
                             task.result()
+                    if wait_reconnect_task not in done:
+                        break
+                    self._reconnect_event.clear()
                 finally:
                     await utils.aio.gracefully_cancel(
                         process_stream_task, wait_reconnect_task
                     )
-            finally:
-                if not self._reconnect_event.is_set():
-                    break
-                self._reconnect_event.clear()
+                    should_stop.set()
+            except DeadlineExceeded:
+                raise APITimeoutError()
+            except GoogleAPICallError as e:
+                raise APIStatusError(
+                    e.message,
+                    status_code=e.code or -1,
+                )
+            except Exception as e:
+                raise APIConnectionError() from e
 
 
 def _recognize_response_to_speech_event(

livekit/plugins/google/version.py

@@ -12,4 +12,4 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-__version__ = "0.9.0"
+__version__ = "0.9.1"

{livekit_plugins_google-0.9.0.dist-info → livekit_plugins_google-0.9.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
-Metadata-Version: 2.1
+Metadata-Version: 2.2
 Name: livekit-plugins-google
-Version: 0.9.0
+Version: 0.9.1
 Summary: Agent Framework plugin for services from Google Cloud
 Home-page: https://github.com/livekit/agents
 License: Apache-2.0
@@ -24,6 +24,16 @@ Requires-Dist: google-cloud-speech<3,>=2
 Requires-Dist: google-cloud-texttospeech<3,>=2
 Requires-Dist: google-genai>=0.3.0
 Requires-Dist: livekit-agents>=0.12.3
+Dynamic: classifier
+Dynamic: description
+Dynamic: description-content-type
+Dynamic: home-page
+Dynamic: keywords
+Dynamic: license
+Dynamic: project-url
+Dynamic: requires-dist
+Dynamic: requires-python
+Dynamic: summary
 
 # LiveKit Plugins Google
 

{livekit_plugins_google-0.9.0.dist-info → livekit_plugins_google-0.9.1.dist-info}/RECORD

@@ -2,14 +2,14 @@ livekit/plugins/google/__init__.py,sha256=TY-5FwEX4Vs7GLO1wSegIxC5W4UPkHBthlr-__
 livekit/plugins/google/log.py,sha256=GI3YWN5YzrafnUccljzPRS_ZALkMNk1i21IRnTl2vNA,69
 livekit/plugins/google/models.py,sha256=cBXhZGY9bFaSCyL9VeSng9wsxhf3peJi3AUYBKV-8GQ,1343
 livekit/plugins/google/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-livekit/plugins/google/stt.py,sha256=SfmKgQotIVzk9-Hipo1X5cnLQG4uXLniTUoyM3IynwA,18712
+livekit/plugins/google/stt.py,sha256=E5kXPbicH4FEXBjyBzfqQWA-nPhKkojzcc-cbtWdmNs,21088
 livekit/plugins/google/tts.py,sha256=95qXCigVQYWNbcN3pIKBpIah4b31U_MWtXv5Ji0AMc4,9229
-livekit/plugins/google/version.py,sha256=onRKrcQ35NZG4oEg_95WGeTytHh_6VVAlQKAZhwiEe4,600
+livekit/plugins/google/version.py,sha256=4GcbYy7J7gvPMEA4wlPB0BJqg8CjF7HRVjQ-i1EH7M8,600
 livekit/plugins/google/beta/__init__.py,sha256=AxRYc7NGG62Tv1MmcZVCDHNvlhbC86hM-_yP01Qb28k,47
 livekit/plugins/google/beta/realtime/__init__.py,sha256=XnJpNIN6NRm7Y4hH2RNA8Xt-tTmkZEKCs_zzU3_koBI,251
 livekit/plugins/google/beta/realtime/api_proto.py,sha256=IHYBryuzpfGQD86Twlfq6qxrBhFHptf_IvOk36Wxo1M,2156
-livekit/plugins/google/beta/realtime/realtime_api.py,sha256=OxrbWnUOT_oFdrMruvLPHgEoXlOr6M5oGym9b2Iqz48,15958
-livekit_plugins_google-0.9.0.dist-info/METADATA,sha256=tB70OQMa7JtWLqRi1TMDUpv4y0TZEk0L609BN6y0x48,1841
-livekit_plugins_google-0.9.0.dist-info/WHEEL,sha256=PZUExdf71Ui_so67QXpySuHtCi3-J3wvF4ORK6k_S8U,91
-livekit_plugins_google-0.9.0.dist-info/top_level.txt,sha256=OoDok3xUmXbZRvOrfvvXB-Juu4DX79dlq188E19YHoo,8
-livekit_plugins_google-0.9.0.dist-info/RECORD,,
+livekit/plugins/google/beta/realtime/realtime_api.py,sha256=YUEf3iR9dIctnXRqev_qKSBM_plqcYKudodFO8nADJY,15966
+livekit_plugins_google-0.9.1.dist-info/METADATA,sha256=y5d0OEdbkoGk0IPGURiDZbt6e6sWhsxOU2cioNrPu7w,2056
+livekit_plugins_google-0.9.1.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
+livekit_plugins_google-0.9.1.dist-info/top_level.txt,sha256=OoDok3xUmXbZRvOrfvvXB-Juu4DX79dlq188E19YHoo,8
+livekit_plugins_google-0.9.1.dist-info/RECORD,,

{livekit_plugins_google-0.9.0.dist-info → livekit_plugins_google-0.9.1.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (75.6.0)
+Generator: setuptools (75.8.0)
 Root-Is-Purelib: true
 Tag: py3-none-any