livekit-plugins-google 1.2.13__tar.gz → 1.2.15__tar.gz

{livekit_plugins_google-1.2.13 → livekit_plugins_google-1.2.15}/.gitignore

@@ -172,4 +172,8 @@ pyrightconfig.json
 docs/
 
 # Database files
-*.db
+*.db
+
+
+# Examples for development
+examples/dev/*

{livekit_plugins_google-1.2.13 → livekit_plugins_google-1.2.15}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: livekit-plugins-google
-Version: 1.2.13
+Version: 1.2.15
 Summary: Agent Framework plugin for services from Google Cloud
 Project-URL: Documentation, https://docs.livekit.io
 Project-URL: Website, https://livekit.io/
@@ -22,7 +22,7 @@ Requires-Dist: google-auth<3,>=2
 Requires-Dist: google-cloud-speech<3,>=2
 Requires-Dist: google-cloud-texttospeech<3,>=2.27
 Requires-Dist: google-genai>=v1.23.0
-Requires-Dist: livekit-agents>=1.2.13
+Requires-Dist: livekit-agents>=1.2.15
 Description-Content-Type: text/markdown
 
 # Google AI plugin for LiveKit Agents

{livekit_plugins_google-1.2.13 → livekit_plugins_google-1.2.15}/livekit/plugins/google/__init__.py

@@ -19,14 +19,14 @@ Supports Gemini, Cloud Speech-to-Text, and Cloud Text-to-Speech.
 See https://docs.livekit.io/agents/integrations/stt/google/ for more information.
 """
 
-from . import beta
+from . import beta, realtime
 from .llm import LLM
 from .stt import STT, SpeechStream
 from .tools import _LLMTool
 from .tts import TTS
 from .version import __version__
 
-__all__ = ["STT", "TTS", "SpeechStream", "__version__", "beta", "LLM", "_LLMTool"]
+__all__ = ["STT", "TTS", "realtime", "SpeechStream", "__version__", "beta", "LLM", "_LLMTool"]
 from livekit.agents import Plugin
 
 from .log import logger

{livekit_plugins_google-1.2.13 → livekit_plugins_google-1.2.15}/livekit/plugins/google/beta/__init__.py

@@ -1,7 +1,7 @@
-from . import realtime
+from .. import realtime
 from .gemini_tts import TTS as GeminiTTS
 
-__all__ = ["realtime", "GeminiTTS"]
+__all__ = ["GeminiTTS", "realtime"]
 
 # Cleanup docs of unexported modules
 _module = dir()

{livekit_plugins_google-1.2.13 → livekit_plugins_google-1.2.15}/livekit/plugins/google/models.py

@@ -10,6 +10,7 @@ SpeechModels = Literal[
     "medical_conversation",
     "chirp",
     "chirp_2",
+    "chirp_3",
     "latest_long",
     "latest_short",
 ]

{livekit_plugins_google-1.2.13/livekit/plugins/google/beta → livekit_plugins_google-1.2.15/livekit/plugins/google}/realtime/api_proto.py

@@ -10,11 +10,43 @@ LiveAPIModels = Literal[
     # models supported on Gemini API
     "gemini-2.0-flash-live-001",
     "gemini-live-2.5-flash-preview",
+    "gemini-2.5-flash-native-audio-preview-09-2025",
     "gemini-2.5-flash-preview-native-audio-dialog",
     "gemini-2.5-flash-exp-native-audio-thinking-dialog",
 ]
 
-Voice = Literal["Puck", "Charon", "Kore", "Fenrir", "Aoede", "Leda", "Orus", "Zephyr"]
+Voice = Literal[
+    "Achernar",
+    "Achird",
+    "Algenib",
+    "Algieba",
+    "Alnilam",
+    "Aoede",
+    "Autonoe",
+    "Callirrhoe",
+    "Charon",
+    "Despina",
+    "Enceladus",
+    "Erinome",
+    "Fenrir",
+    "Gacrux",
+    "Iapetus",
+    "Kore",
+    "Laomedeia",
+    "Leda",
+    "Orus",
+    "Pulcherrima",
+    "Puck",
+    "Rasalgethi",
+    "Sadachbia",
+    "Sadaltager",
+    "Schedar",
+    "Sulafat",
+    "Umbriel",
+    "Vindemiatrix",
+    "Zephyr",
+    "Zubenelgenubi",
+]
 
 
 ClientEvents = Union[

{livekit_plugins_google-1.2.13/livekit/plugins/google/beta → livekit_plugins_google-1.2.15/livekit/plugins/google}/realtime/realtime_api.py

@@ -23,11 +23,11 @@ from livekit.agents.types import (
     NotGivenOr,
 )
 from livekit.agents.utils import audio as audio_utils, images, is_given
-from livekit.plugins.google.beta.realtime.api_proto import ClientEvents, LiveAPIModels, Voice
+from livekit.plugins.google.realtime.api_proto import ClientEvents, LiveAPIModels, Voice
 
-from ...log import logger
-from ...tools import _LLMTool
-from ...utils import create_tools_config, get_tool_results_for_realtime, to_fnc_ctx
+from ..log import logger
+from ..tools import _LLMTool
+from ..utils import create_tools_config, get_tool_results_for_realtime, to_fnc_ctx
 
 INPUT_AUDIO_SAMPLE_RATE = 16000
 INPUT_AUDIO_CHANNELS = 1
@@ -78,6 +78,7 @@ class _RealtimeOptions:
     gemini_tools: NotGivenOr[list[_LLMTool]] = NOT_GIVEN
     tool_behavior: NotGivenOr[types.Behavior] = NOT_GIVEN
     tool_response_scheduling: NotGivenOr[types.FunctionResponseScheduling] = NOT_GIVEN
+    thinking_config: NotGivenOr[types.ThinkingConfig] = NOT_GIVEN
 
 
 @dataclass
@@ -144,6 +145,7 @@ class RealtimeModel(llm.RealtimeModel):
         conn_options: APIConnectOptions = DEFAULT_API_CONNECT_OPTIONS,
         http_options: NotGivenOr[types.HttpOptions] = NOT_GIVEN,
         _gemini_tools: NotGivenOr[list[_LLMTool]] = NOT_GIVEN,
+        thinking_config: NotGivenOr[types.ThinkingConfig] = NOT_GIVEN,
     ) -> None:
         """
         Initializes a RealtimeModel instance for interacting with Google's Realtime API.
@@ -180,6 +182,7 @@ class RealtimeModel(llm.RealtimeModel):
             context_window_compression (ContextWindowCompressionConfig, optional): The configuration for context window compression. Defaults to None.
             tool_behavior (Behavior, optional): The behavior for tool call. Default behavior is BLOCK in Gemini Realtime API.
             tool_response_scheduling (FunctionResponseScheduling, optional): The scheduling for tool response. Default scheduling is WHEN_IDLE.
+            thinking_config (ThinkingConfig, optional): Native audio thinking configuration.
             conn_options (APIConnectOptions, optional): The configuration for the API connection. Defaults to DEFAULT_API_CONNECT_OPTIONS.
             _gemini_tools (list[LLMTool], optional): Gemini-specific tools to use for the session. This parameter is experimental and may change.
 
@@ -274,6 +277,7 @@ class RealtimeModel(llm.RealtimeModel):
             tool_behavior=tool_behavior,
             conn_options=conn_options,
             http_options=http_options,
+            thinking_config=thinking_config,
         )
 
         self._sessions = weakref.WeakSet[RealtimeSession]()
@@ -510,7 +514,12 @@ class RealtimeSession(llm.RealtimeSession):
         for f in self._resample_audio(frame):
             for nf in self._bstream.write(f.data.tobytes()):
                 realtime_input = types.LiveClientRealtimeInput(
-                    media_chunks=[types.Blob(data=nf.data.tobytes(), mime_type="audio/pcm")]
+                    media_chunks=[
+                        types.Blob(
+                            data=nf.data.tobytes(),
+                            mime_type=f"audio/pcm;rate={INPUT_AUDIO_SAMPLE_RATE}",
+                        )
+                    ]
                 )
                 self._send_client_event(realtime_input)
 
@@ -814,6 +823,9 @@ class RealtimeSession(llm.RealtimeSession):
                 frequency_penalty=self._opts.frequency_penalty
                 if is_given(self._opts.frequency_penalty)
                 else None,
+                thinking_config=self._opts.thinking_config
+                if is_given(self._opts.thinking_config)
+                else None,
             ),
             system_instruction=types.Content(parts=[types.Part(text=self._opts.instructions)])
             if is_given(self._opts.instructions)

{livekit_plugins_google-1.2.13 → livekit_plugins_google-1.2.15}/livekit/plugins/google/stt.py

@@ -618,17 +618,28 @@ def _streaming_recognize_response_to_speech_data(
 ) -> stt.SpeechData | None:
     text = ""
     confidence = 0.0
+    final_result = None
     for result in resp.results:
         if len(result.alternatives) == 0:
             continue
-        text += result.alternatives[0].transcript
-        confidence += result.alternatives[0].confidence
-
-    confidence /= len(resp.results)
-    lg = resp.results[0].language_code
+        else:
+            if result.is_final:
+                final_result = result
+                break
+            else:
+                text += result.alternatives[0].transcript
+                confidence += result.alternatives[0].confidence
+
+    if final_result is not None:
+        text = final_result.alternatives[0].transcript
+        confidence = final_result.alternatives[0].confidence
+        lg = final_result.language_code
+    else:
+        confidence /= len(resp.results)
+        if confidence < min_confidence_threshold:
+            return None
+        lg = resp.results[0].language_code
 
-    if confidence < min_confidence_threshold:
-        return None
     if text == "":
         return None
 

{livekit_plugins_google-1.2.13 → livekit_plugins_google-1.2.15}/livekit/plugins/google/version.py

@@ -12,4 +12,4 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-__version__ = "1.2.13"
+__version__ = "1.2.15"

{livekit_plugins_google-1.2.13 → livekit_plugins_google-1.2.15}/pyproject.toml

@@ -27,7 +27,7 @@ dependencies = [
     "google-cloud-speech >= 2, < 3",
     "google-cloud-texttospeech >= 2.27, < 3",
     "google-genai >= v1.23.0",
-    "livekit-agents>=1.2.13",
+    "livekit-agents>=1.2.15",
 ]
 
 [project.urls]