sarvamai 0.1.19a0__tar.gz → 0.1.19a1__tar.gz

{sarvamai-0.1.19a0 → sarvamai-0.1.19a1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: sarvamai
-Version: 0.1.19a0
+Version: 0.1.19a1
 Summary: 
 Requires-Python: >=3.8,<4.0
 Classifier: Intended Audience :: Developers

{sarvamai-0.1.19a0 → sarvamai-0.1.19a1}/pyproject.toml

@@ -3,7 +3,7 @@ name = "sarvamai"
 
 [tool.poetry]
 name = "sarvamai"
-version = "0.1.19a0"
+version = "0.1.19a1"
 description = ""
 readme = "README.md"
 authors = []

{sarvamai-0.1.19a0 → sarvamai-0.1.19a1}/src/sarvamai/core/client_wrapper.py

@@ -23,10 +23,10 @@ class BaseClientWrapper:
 
     def get_headers(self) -> typing.Dict[str, str]:
         headers: typing.Dict[str, str] = {
-            "User-Agent": "sarvamai/0.1.19a0",
+            "User-Agent": "sarvamai/0.1.19a1",
             "X-Fern-Language": "Python",
             "X-Fern-SDK-Name": "sarvamai",
-            "X-Fern-SDK-Version": "0.1.19a0",
+            "X-Fern-SDK-Version": "0.1.19a1",
             **(self.get_custom_headers() or {}),
         }
         headers["api-subscription-key"] = self.api_subscription_key

{sarvamai-0.1.19a0 → sarvamai-0.1.19a1}/src/sarvamai/speech_to_text_streaming/socket_client.py

@@ -9,7 +9,10 @@ from ..core.events import EventEmitterMixin, EventType
 from ..core.pydantic_utilities import parse_obj_as
 from ..types.audio_data import AudioData
 from ..types.audio_message import AudioMessage
-from ..types.speech_to_text_streaming_response import SpeechToTextStreamingResponse
+from ..types.speech_to_text_streaming_response import (
+    SpeechToTextStreamingResponse,
+)
+from ..types.stt_flush_signal import SttFlushSignal
 
 SpeechToTextStreamingSocketClientResponse = typing.Union[SpeechToTextStreamingResponse]
 
@@ -22,7 +25,9 @@ class AsyncSpeechToTextStreamingSocketClient(EventEmitterMixin):
     async def __aiter__(self):
         async for message in self._websocket:
             message = json.loads(message) if isinstance(message, str) else message
-            yield parse_obj_as(SpeechToTextStreamingSocketClientResponse, message)  # type: ignore
+            yield parse_obj_as(
+                SpeechToTextStreamingSocketClientResponse, message
+            )  # type: ignore
 
     async def start_listening(self):
         """
@@ -37,8 +42,14 @@ class AsyncSpeechToTextStreamingSocketClient(EventEmitterMixin):
         self._emit(EventType.OPEN, None)
         try:
             async for raw_message in self._websocket:
-                raw_message = json.loads(raw_message) if isinstance(raw_message, str) else raw_message
-                parsed = parse_obj_as(SpeechToTextStreamingSocketClientResponse, raw_message)  # type: ignore
+                raw_message = (
+                    json.loads(raw_message)
+                    if isinstance(raw_message, str)
+                    else raw_message
+                )
+                parsed = parse_obj_as(
+                    SpeechToTextStreamingSocketClientResponse, raw_message
+                )  # type: ignore
                 self._emit(EventType.MESSAGE, parsed)
         except websockets.WebSocketException as exc:
             self._emit(EventType.ERROR, exc)
@@ -54,10 +65,26 @@ class AsyncSpeechToTextStreamingSocketClient(EventEmitterMixin):
         """
 
         return await self._send_speech_to_text_streaming_audio_message(
-            message=AudioMessage(audio=AudioData(data=audio, sample_rate=sample_rate, encoding=encoding))
+            message=AudioMessage(
+                audio=AudioData(data=audio, sample_rate=sample_rate, encoding=encoding)
+            )
         )
 
-    async def _send_speech_to_text_streaming_audio_message(self, message: AudioMessage) -> None:
+    async def flush(self, reason: typing.Optional[str] = None) -> None:
+        """
+        Signal to flush the audio buffer and force finalize partial transcriptions.
+        Use this to force processing of any remaining audio that hasn't been
+        transcribed yet.
+
+        :param reason: Optional reason for flushing (e.g., "end_of_segment",
+                      "manual_flush")
+        """
+        message = SttFlushSignal(reason=reason)
+        await self._send_model(message)
+
+    async def _send_speech_to_text_streaming_audio_message(
+        self, message: AudioMessage
+    ) -> None:
         """
         Send a message to the websocket connection.
         The message will be sent as a AudioMessage.
@@ -70,7 +97,9 @@ class AsyncSpeechToTextStreamingSocketClient(EventEmitterMixin):
         """
         data = await self._websocket.recv()
         data = json.loads(data) if isinstance(data, str) else data
-        return parse_obj_as(SpeechToTextStreamingSocketClientResponse, data)  # type: ignore
+        return parse_obj_as(
+            SpeechToTextStreamingSocketClientResponse, data
+        )  # type: ignore
 
     async def _send(self, data: typing.Any) -> None:
         """
@@ -95,7 +124,9 @@ class SpeechToTextStreamingSocketClient(EventEmitterMixin):
     def __iter__(self):
         for message in self._websocket:
             message = json.loads(message) if isinstance(message, str) else message
-            yield parse_obj_as(SpeechToTextStreamingSocketClientResponse, message)  # type: ignore
+            yield parse_obj_as(
+                SpeechToTextStreamingSocketClientResponse, message
+            )  # type: ignore
 
     def start_listening(self):
         """
@@ -110,8 +141,14 @@ class SpeechToTextStreamingSocketClient(EventEmitterMixin):
         self._emit(EventType.OPEN, None)
         try:
             for raw_message in self._websocket:
-                raw_message = json.loads(raw_message) if isinstance(raw_message, str) else raw_message
-                parsed = parse_obj_as(SpeechToTextStreamingSocketClientResponse, raw_message)  # type: ignore
+                raw_message = (
+                    json.loads(raw_message)
+                    if isinstance(raw_message, str)
+                    else raw_message
+                )
+                parsed = parse_obj_as(
+                    SpeechToTextStreamingSocketClientResponse, raw_message
+                )  # type: ignore
                 self._emit(EventType.MESSAGE, parsed)
         except websockets.WebSocketException as exc:
             self._emit(EventType.ERROR, exc)
@@ -126,18 +163,36 @@ class SpeechToTextStreamingSocketClient(EventEmitterMixin):
         :param sample_rate (Optional): Audio sample rate in Hz (default is 16000)
         """
         return self._send_speech_to_text_streaming_audio_message(
-            message=AudioMessage(audio=AudioData(data=audio, sample_rate=sample_rate, encoding=encoding))
+            message=AudioMessage(
+                audio=AudioData(data=audio, sample_rate=sample_rate, encoding=encoding)
+            )
         )
 
+    def flush(self, reason: typing.Optional[str] = None) -> None:
+        """
+        Signal to flush the audio buffer and force finalize partial transcriptions.
+        Use this to force processing of any remaining audio that hasn't been
+        transcribed yet.
+
+        :param reason: Optional reason for flushing (e.g., "end_of_segment",
+                      "manual_flush")
+        """
+        message = SttFlushSignal(reason=reason)
+        self._send_model(message)
+
     def recv(self) -> SpeechToTextStreamingSocketClientResponse:
         """
         Receive a message from the websocket connection.
         """
         data = self._websocket.recv()
         data = json.loads(data) if isinstance(data, str) else data
-        return parse_obj_as(SpeechToTextStreamingSocketClientResponse, data)  # type: ignore
+        return parse_obj_as(
+            SpeechToTextStreamingSocketClientResponse, data
+        )  # type: ignore
 
-    def _send_speech_to_text_streaming_audio_message(self, message: AudioMessage) -> None:
+    def _send_speech_to_text_streaming_audio_message(
+        self, message: AudioMessage
+    ) -> None:
         """
         Send a message to the websocket connection.
         The message will be sent as a AudioMessage.