livekit-plugins-aws 1.2.4__py3-none-any.whl → 1.3.2__py3-none-any.whl

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__()
@@ -111,12 +117,18 @@ 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
     def model(self) -> str:
         return self._opts.model
 
+    @property
+    def provider(self) -> str:
+        return "AWS Bedrock"
+
     def chat(
         self,
         *,
@@ -140,7 +152,11 @@ 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 +178,12 @@ 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/models.py

@@ -46,3 +46,4 @@ TTSLanguages = Literal[
 ]
 
 TTSEncoding = Literal["mp3"]
+TTSTextType = Literal["text", "ssml"]

livekit/plugins/aws/stt.py

@@ -16,10 +16,11 @@ import asyncio
 import os
 from dataclasses import dataclass
 
-from amazon_transcribe.auth import AwsCrtCredentialResolver
+from amazon_transcribe.auth import AwsCrtCredentialResolver, CredentialResolver, Credentials
 from amazon_transcribe.client import TranscribeStreamingClient
 from amazon_transcribe.exceptions import BadRequestException
 from amazon_transcribe.model import Result, StartStreamTranscriptionEventStream, TranscriptEvent
+from awscrt.auth import AwsCredentialsProvider  # type: ignore[import-untyped]
 
 from livekit import rtc
 from livekit.agents import (
@@ -71,6 +72,7 @@ class STT(stt.STT):
         enable_partial_results_stabilization: NotGivenOr[bool] = NOT_GIVEN,
         partial_results_stability: NotGivenOr[str] = NOT_GIVEN,
         language_model_name: NotGivenOr[str] = NOT_GIVEN,
+        credentials: NotGivenOr[Credentials] = NOT_GIVEN,
     ):
         super().__init__(capabilities=stt.STTCapabilities(streaming=True, interim_results=True))
 
@@ -94,6 +96,20 @@ class STT(stt.STT):
             region=region,
         )
 
+        self._credentials = credentials if is_given(credentials) else None
+
+    @property
+    def model(self) -> str:
+        return (
+            self._config.language_model_name
+            if is_given(self._config.language_model_name)
+            else "unknown"
+        )
+
+    @property
+    def provider(self) -> str:
+        return "Amazon Transcribe"
+
     async def aclose(self) -> None:
         await super().aclose()
 
@@ -112,7 +128,9 @@ class STT(stt.STT):
         language: NotGivenOr[str] = NOT_GIVEN,
         conn_options: APIConnectOptions = DEFAULT_API_CONNECT_OPTIONS,
     ) -> SpeechStream:
-        return SpeechStream(stt=self, conn_options=conn_options, opts=self._config)
+        return SpeechStream(
+            stt=self, conn_options=conn_options, opts=self._config, credentials=self._credentials
+        )
 
 
 class SpeechStream(stt.SpeechStream):
@@ -121,15 +139,37 @@ class SpeechStream(stt.SpeechStream):
         stt: STT,
         opts: STTOptions,
         conn_options: APIConnectOptions = DEFAULT_API_CONNECT_OPTIONS,
+        credentials: Credentials | None = None,
     ) -> None:
         super().__init__(stt=stt, conn_options=conn_options, sample_rate=opts.sample_rate)
         self._opts = opts
+        self._credentials = credentials
+
+    def _credential_resolver(self) -> CredentialResolver:
+        if self._credentials is None:
+            return AwsCrtCredentialResolver(None)  # type: ignore
+
+        credentials = self._credentials
+
+        class CustomAwsCrtCredentialResolver(CredentialResolver):
+            def __init__(self) -> None:
+                self._crt_resolver = AwsCredentialsProvider.new_static(
+                    credentials.access_key_id,
+                    credentials.secret_access_key,
+                    credentials.session_token,
+                )
+
+            async def get_credentials(self) -> Credentials | None:
+                credentials = await asyncio.wrap_future(self._crt_resolver.get_credentials())
+                return credentials  # type: ignore[no-any-return]
+
+        return CustomAwsCrtCredentialResolver()
 
     async def _run(self) -> None:
         while True:
             client = TranscribeStreamingClient(
                 region=self._opts.region,
-                credential_resolver=AwsCrtCredentialResolver(None),  # type: ignore
+                credential_resolver=self._credential_resolver(),
             )
 
             live_config = {
@@ -192,7 +232,7 @@ class SpeechStream(stt.SpeechStream):
                     self._event_ch.send_nowait(
                         stt.SpeechEvent(
                             type=stt.SpeechEventType.INTERIM_TRANSCRIPT,
-                            alternatives=[_streaming_recognize_response_to_speech_data(resp)],
+                            alternatives=[self._streaming_recognize_response_to_speech_data(resp)],
                         )
                     )
 
@@ -200,20 +240,22 @@ class SpeechStream(stt.SpeechStream):
                     self._event_ch.send_nowait(
                         stt.SpeechEvent(
                             type=stt.SpeechEventType.FINAL_TRANSCRIPT,
-                            alternatives=[_streaming_recognize_response_to_speech_data(resp)],
+                            alternatives=[self._streaming_recognize_response_to_speech_data(resp)],
                         )
                     )
 
             if not resp.is_partial:
                 self._event_ch.send_nowait(stt.SpeechEvent(type=stt.SpeechEventType.END_OF_SPEECH))
 
+    def _streaming_recognize_response_to_speech_data(self, resp: Result) -> stt.SpeechData:
+        confidence = 0.0
+        if resp.alternatives and (items := resp.alternatives[0].items):
+            confidence = items[0].confidence or 0.0
 
-def _streaming_recognize_response_to_speech_data(resp: Result) -> stt.SpeechData:
-    data = stt.SpeechData(
-        language="en-US",
-        start_time=resp.start_time if resp.start_time else 0.0,
-        end_time=resp.end_time if resp.end_time else 0.0,
-        text=resp.alternatives[0].transcript if resp.alternatives else "",
-    )
-
-    return data
+        return stt.SpeechData(
+            language=resp.language_code or self._opts.language,
+            start_time=resp.start_time if resp.start_time is not None else 0.0,
+            end_time=resp.end_time if resp.end_time is not None else 0.0,
+            text=resp.alternatives[0].transcript if resp.alternatives else "",
+            confidence=confidence,
+        )

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
@@ -32,11 +33,12 @@ from livekit.agents.types import (
 )
 from livekit.agents.utils import is_given
 
-from .models import TTSLanguages, TTSSpeechEngine
+from .models import TTSLanguages, TTSSpeechEngine, TTSTextType
 from .utils import _strip_nones
 
 DEFAULT_SPEECH_ENGINE: TTSSpeechEngine = "generative"
 DEFAULT_VOICE = "Ruth"
+DEFAULT_TEXT_TYPE: TTSTextType = "text"
 
 
 @dataclass
@@ -47,6 +49,7 @@ class _TTSOptions:
     region: str | None
     sample_rate: int
     language: TTSLanguages | str | None
+    text_type: TTSTextType
 
 
 class TTS(tts.TTS):
@@ -56,6 +59,7 @@ class TTS(tts.TTS):
         voice: str = "Ruth",
         language: NotGivenOr[TTSLanguages | str] = NOT_GIVEN,
         speech_engine: TTSSpeechEngine = "generative",
+        text_type: TTSTextType = "text",
         sample_rate: int = 16000,
         region: str | None = None,
         api_key: str | None = None,
@@ -71,10 +75,11 @@ class TTS(tts.TTS):
         See https://docs.aws.amazon.com/polly/latest/dg/API_SynthesizeSpeech.html for more details on the the AWS Polly TTS.
 
         Args:
-            Voice (TTSModels, optional): Voice ID to use for the synthesis. Defaults to "Ruth".
-            language (TTS_LANGUAGE, optional): language code for the Synthesize Speech request. This is only necessary if using a bilingual voice, such as Aditi, which can be used for either Indian English (en-IN) or Hindi (hi-IN).
+            voice (TTSModels, optional): Voice ID to use for the synthesis. Defaults to "Ruth".
+            language (TTSLanguages, optional): language code for the Synthesize Speech request. This is only necessary if using a bilingual voice, such as Aditi, which can be used for either Indian English (en-IN) or Hindi (hi-IN).
+            speech_engine(TTSSpeechEngine, optional): The engine to use for the synthesis. Defaults to "generative".
+            text_type(TTSTextType, optional): Type of text to synthesize. Use "ssml" for SSML-enhanced text. Defaults to "text".
             sample_rate(int, optional): The audio frequency specified in Hz. Defaults to 16000.
-            speech_engine(TTS_SPEECH_ENGINE, optional): The engine to use for the synthesis. Defaults to "generative".
             region(str, optional): The region to use for the synthesis. Defaults to "us-east-1".
             api_key(str, optional): AWS access key id.
             api_secret(str, optional): AWS secret access key.
@@ -96,16 +101,42 @@ class TTS(tts.TTS):
         self._opts = _TTSOptions(
             voice=voice,
             speech_engine=speech_engine,
+            text_type=text_type,
             region=region or None,
             language=language or None,
             sample_rate=sample_rate,
         )
 
+    @property
+    def model(self) -> str:
+        return self._opts.speech_engine
+
+    @property
+    def provider(self) -> str:
+        return "Amazon Polly"
+
     def synthesize(
         self, text: str, *, conn_options: APIConnectOptions = DEFAULT_API_CONNECT_OPTIONS
     ) -> 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__(
@@ -130,7 +161,7 @@ class ChunkedStream(tts.ChunkedStream):
                             "OutputFormat": "mp3",
                             "Engine": self._opts.speech_engine,
                             "VoiceId": self._opts.voice,
-                            "TextType": "text",
+                            "TextType": self._opts.text_type,
                             "SampleRate": str(self._opts.sample_rate),
                             "LanguageCode": self._opts.language,
                         }

livekit/plugins/aws/version.py

@@ -12,4 +12,4 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-__version__ = "1.2.4"
+__version__ = "1.3.2"

{livekit_plugins_aws-1.2.4.dist-info → livekit_plugins_aws-1.3.2.dist-info}/METADATA

@@ -1,13 +1,13 @@
 Metadata-Version: 2.4
 Name: livekit-plugins-aws
-Version: 1.2.4
+Version: 1.3.2
 Summary: LiveKit Agents Plugin for services from AWS
 Project-URL: Documentation, https://docs.livekit.io
 Project-URL: Website, https://livekit.io/
 Project-URL: Source, https://github.com/livekit/agents
 Author-email: LiveKit <hello@livekit.io>
 License-Expression: Apache-2.0
-Keywords: audio,aws,livekit,nova,realtime,sonic,video,webrtc
+Keywords: ai,audio,aws,livekit,nova,realtime,sonic,video,voice
 Classifier: Intended Audience :: Developers
 Classifier: License :: OSI Approved :: Apache Software License
 Classifier: Programming Language :: Python :: 3
@@ -19,10 +19,11 @@ Classifier: Topic :: Multimedia :: Video
 Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
 Requires-Python: >=3.9.0
 Requires-Dist: aioboto3>=14.1.0
-Requires-Dist: amazon-transcribe>=0.6.2
-Requires-Dist: livekit-agents>=1.2.4
+Requires-Dist: amazon-transcribe>=0.6.4
+Requires-Dist: livekit-agents>=1.3.2
 Provides-Extra: realtime
 Requires-Dist: aws-sdk-bedrock-runtime==0.0.2; (python_version >= '3.12') and extra == 'realtime'
+Requires-Dist: aws-sdk-signers==0.0.3; (python_version >= '3.12') and extra == 'realtime'
 Requires-Dist: boto3>1.35.10; extra == 'realtime'
 Description-Content-Type: text/markdown
 

{livekit_plugins_aws-1.2.4.dist-info → livekit_plugins_aws-1.3.2.dist-info}/RECORD

@@ -1,17 +1,17 @@
 livekit/plugins/aws/__init__.py,sha256=dCZISj1yZG0WZTojk3sU-Ub4PK1ThCVhamrl9k_NbBw,2047
-livekit/plugins/aws/llm.py,sha256=SUPWhJTbQ6HZJEK7WYUADDo2BJZJl2EaRvfG05IobzU,12150
+livekit/plugins/aws/llm.py,sha256=9adQTcg3hJA6XTw4xaRjCIKkxedbzpNBBW0Yub9pkhA,13001
 livekit/plugins/aws/log.py,sha256=S5ICcsnwshZhMG0HPmc_lI3mtHmcY4oQMJBsnnho-bM,289
-livekit/plugins/aws/models.py,sha256=ztI9n3r8OHGlQewWXiyQP3SaavFKHkw-VVuK9Xz4K1I,696
+livekit/plugins/aws/models.py,sha256=J4yzik9sR68RPZpR1ubRQ9hdn14D9IwA3KaRvAf5tAE,734
 livekit/plugins/aws/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-livekit/plugins/aws/stt.py,sha256=zlUrIVVYsSMhtVzXVxHTypW_E2YpMD9EAqvW3lZYj1c,8888
-livekit/plugins/aws/tts.py,sha256=T5dVpTuIuzQimYNnkfXi5dRLmRldWySL4IcbkXjmJLM,6083
+livekit/plugins/aws/stt.py,sha256=WXE25wXwCgKeTcoMe3AH9EUeomvvRtPpOTZ5JAfOUxk,10629
+livekit/plugins/aws/tts.py,sha256=oav-XWf9ysVGCmERWej6BgACu8vsLbRo9vFGpo9N6Ec,7184
 livekit/plugins/aws/utils.py,sha256=nA5Ua1f4T-25Loar6EvlrKTXI9N-zpTIH7cdQkwGyGI,1518
-livekit/plugins/aws/version.py,sha256=RNHljWBeimzzNkMMbX3wjVENjAQ3D1xYW3wp8ijSE3s,600
+livekit/plugins/aws/version.py,sha256=fSSiY4SPcmDoVdQmTTNkDwxlIEOPaqT_xdow1m-W9JQ,600
 livekit/plugins/aws/experimental/realtime/__init__.py,sha256=mm_TGZc9QAWSO-VOO3PdE8Y5R6xlWckXRZuiFUIHa-Q,287
 livekit/plugins/aws/experimental/realtime/events.py,sha256=ltdGEipE3ZOkjn7K6rKN6WSCUPJkVg-S88mUmQ_V00s,15981
 livekit/plugins/aws/experimental/realtime/pretty_printer.py,sha256=KN7KPrfQu8cU7ff34vFAtfrd1umUSTVNKXQU7D8AMiM,1442
-livekit/plugins/aws/experimental/realtime/realtime_model.py,sha256=JkFv4LnlME17v-yXVvZiFdFyKHZBkKlOffmbUAd7qYw,60403
+livekit/plugins/aws/experimental/realtime/realtime_model.py,sha256=ksdw7X-wm5wiDoCur9srYTRV2eVadHOjAIIanNS9dUo,64568
 livekit/plugins/aws/experimental/realtime/turn_tracker.py,sha256=bcufaap-coeIYuK3ct1Is9W_UoefGYRmnJu7Mn5DCYU,6002
-livekit_plugins_aws-1.2.4.dist-info/METADATA,sha256=ExAhD6Tb7l1DuATSpAUjHXECCXHEzrcA43oNlmPmrFQ,1989
-livekit_plugins_aws-1.2.4.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-livekit_plugins_aws-1.2.4.dist-info/RECORD,,
+livekit_plugins_aws-1.3.2.dist-info/METADATA,sha256=BdW-6sTdtruSTfvIgdDU_m1VQb2A6E4z9HaNSeVPpNQ,2081
+livekit_plugins_aws-1.3.2.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+livekit_plugins_aws-1.3.2.dist-info/RECORD,,