livekit-plugins-aws 1.2.8__tar.gz → 1.2.10__tar.gz

{livekit_plugins_aws-1.2.8 → livekit_plugins_aws-1.2.10}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: livekit-plugins-aws
-Version: 1.2.8
+Version: 1.2.10
 Summary: LiveKit Agents Plugin for services from AWS
 Project-URL: Documentation, https://docs.livekit.io
 Project-URL: Website, https://livekit.io/
@@ -20,7 +20,7 @@ Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
 Requires-Python: >=3.9.0
 Requires-Dist: aioboto3>=14.1.0
 Requires-Dist: amazon-transcribe>=0.6.4
-Requires-Dist: livekit-agents>=1.2.8
+Requires-Dist: livekit-agents>=1.2.10
 Provides-Extra: realtime
 Requires-Dist: aws-sdk-bedrock-runtime==0.0.2; (python_version >= '3.12') and extra == 'realtime'
 Requires-Dist: boto3>1.35.10; extra == 'realtime'

{livekit_plugins_aws-1.2.8 → livekit_plugins_aws-1.2.10}/livekit/plugins/aws/experimental/realtime/realtime_model.py

@@ -240,6 +240,7 @@ class RealtimeModel(llm.RealtimeModel):
                 user_transcription=True,
                 auto_tool_reply_generation=True,
                 audio_output=True,
+                manual_function_calls=False,
             )
         )
         self.model_id = "amazon.nova-sonic-v1:0"
@@ -272,6 +273,10 @@ class RealtimeModel(llm.RealtimeModel):
         async def aclose(self) -> None:
             pass
 
+    @property
+    def model(self) -> str:
+        return self.model_id
+
 
 class RealtimeSession(  # noqa: F811
     llm.RealtimeSession[Literal["bedrock_server_event_received", "bedrock_client_event_queued"]]
@@ -556,6 +561,7 @@ class RealtimeSession(  # noqa: F811
             message_stream=self._current_generation.message_ch,
             function_stream=self._current_generation.function_ch,
             user_initiated=False,
+            response_id=self._current_generation.response_id,
         )
         self.emit("generation_created", generation_ev)
 
@@ -864,7 +870,8 @@ class RealtimeSession(  # noqa: F811
         output_tokens = event_data["event"]["usageEvent"]["details"]["delta"]["output"]
         # Q: should we be counting per turn or utterance?
         metrics = RealtimeModelMetrics(
-            label=self._realtime_model._label,
+            label=self._realtime_model.label,
+            model=self._realtime_model.model,
             # TODO: pass in the correct request_id
             request_id=event_data["event"]["usageEvent"]["completionId"],
             timestamp=time.monotonic(),

{livekit_plugins_aws-1.2.8 → livekit_plugins_aws-1.2.10}/livekit/plugins/aws/llm.py

@@ -51,6 +51,8 @@ class _LLMOptions:
     max_output_tokens: NotGivenOr[int]
     top_p: NotGivenOr[float]
     additional_request_fields: NotGivenOr[dict[str, Any]]
+    cache_system: bool
+    cache_tools: bool
 
 
 class LLM(llm.LLM):
@@ -66,6 +68,8 @@ class LLM(llm.LLM):
         top_p: NotGivenOr[float] = NOT_GIVEN,
         tool_choice: NotGivenOr[ToolChoice] = NOT_GIVEN,
         additional_request_fields: NotGivenOr[dict[str, Any]] = NOT_GIVEN,
+        cache_system: bool = False,
+        cache_tools: bool = False,
         session: aioboto3.Session | None = None,
     ) -> None:
         """
@@ -87,6 +91,8 @@ class LLM(llm.LLM):
             top_p (float, optional): The nucleus sampling probability for response generation. Defaults to None.
             tool_choice (ToolChoice, optional): Specifies whether to use tools during response generation. Defaults to "auto".
             additional_request_fields (dict[str, Any], optional): Additional request fields to send to the AWS Bedrock Converse API. Defaults to None.
+            cache_system (bool, optional): Caches system messages to reduce token usage. Defaults to False.
+            cache_tools (bool, optional): Caches tool definitions to reduce token usage. Defaults to False.
             session (aioboto3.Session, optional): Optional aioboto3 session to use.
         """  # noqa: E501
         super().__init__()
@@ -111,6 +117,8 @@ class LLM(llm.LLM):
             max_output_tokens=max_output_tokens,
             top_p=top_p,
             additional_request_fields=additional_request_fields,
+            cache_system=cache_system,
+            cache_tools=cache_tools,
         )
 
     @property
@@ -140,7 +148,11 @@ class LLM(llm.LLM):
             if not tools:
                 return None
 
-            tool_config: dict[str, Any] = {"tools": to_fnc_ctx(tools)}
+            tools_list = to_fnc_ctx(tools)
+            if self._opts.cache_tools:
+                tools_list.append({"cachePoint": {"type": "default"}})
+
+            tool_config: dict[str, Any] = {"tools": tools_list}
             tool_choice = (
                 cast(ToolChoice, tool_choice) if is_given(tool_choice) else self._opts.tool_choice
             )
@@ -162,7 +174,12 @@ class LLM(llm.LLM):
         messages, extra_data = chat_ctx.to_provider_format(format="aws")
         opts["messages"] = messages
         if extra_data.system_messages:
-            opts["system"] = [{"text": content} for content in extra_data.system_messages]
+            system_messages: list[dict[str, str | dict]] = [
+                {"text": content} for content in extra_data.system_messages
+            ]
+            if self._opts.cache_system:
+                system_messages.append({"cachePoint": {"type": "default"}})
+            opts["system"] = system_messages
 
         inference_config: dict[str, Any] = {}
         if is_given(self._opts.max_output_tokens):

{livekit_plugins_aws-1.2.8 → livekit_plugins_aws-1.2.10}/livekit/plugins/aws/tts.py

@@ -13,6 +13,7 @@
 from __future__ import annotations
 
 from dataclasses import dataclass, replace
+from typing import cast
 
 import aioboto3  # type: ignore
 import botocore  # type: ignore
@@ -111,6 +112,23 @@ class TTS(tts.TTS):
     ) -> ChunkedStream:
         return ChunkedStream(tts=self, text=text, conn_options=conn_options)
 
+    def update_options(
+        self,
+        *,
+        voice: NotGivenOr[str] = NOT_GIVEN,
+        language: NotGivenOr[str] = NOT_GIVEN,
+        speech_engine: NotGivenOr[TTSSpeechEngine] = NOT_GIVEN,
+        text_type: NotGivenOr[TTSTextType] = NOT_GIVEN,
+    ) -> None:
+        if is_given(voice):
+            self._opts.voice = voice
+        if is_given(language):
+            self._opts.language = language
+        if is_given(speech_engine):
+            self._opts.speech_engine = cast(TTSSpeechEngine, speech_engine)
+        if is_given(text_type):
+            self._opts.text_type = cast(TTSTextType, text_type)
+
 
 class ChunkedStream(tts.ChunkedStream):
     def __init__(

{livekit_plugins_aws-1.2.8 → livekit_plugins_aws-1.2.10}/livekit/plugins/aws/version.py

@@ -12,4 +12,4 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-__version__ = "1.2.8"
+__version__ = '1.2.10'

{livekit_plugins_aws-1.2.8 → livekit_plugins_aws-1.2.10}/pyproject.toml

@@ -23,7 +23,7 @@ classifiers = [
     "Programming Language :: Python :: 3 :: Only",
 ]
 dependencies = [
-    "livekit-agents>=1.2.8",
+    "livekit-agents>=1.2.10",
     "aioboto3>=14.1.0",
     "amazon-transcribe>=0.6.4",
 ]