Merge pull request #50 from speechmatics/feature/force-end-of-utterance

Add force eou

Author: J-Jaywalker

sdk/rt/speechmatics/rt/_async_client.py

@@ -161,6 +161,26 @@ async def stop_session(self) -> None:
         await self._session_done_evt.wait()  # Wait for end of transcript event to indicate we can stop listening
         await self.close()
 
+    async def force_end_of_utterance(self) -> None:
+        """
+    This method sends a ForceEndOfUtterance message to the server to signal
+        the end of an utterance. Forcing end of utterance will cause the final
+        transcript to be sent to the client early.
+
+        Raises:
+            ConnectionError: If the WebSocket connection fails.
+            TranscriptionError: If the server reports an error during teardown.
+            TimeoutError: If the connection or teardown times out.
+
+        Examples:
+            Basic streaming:
+                >>> async with AsyncClient() as client:
+                ...     await client.start_session()
+                ...     await client.send_audio(frame)
+                ...     await client.force_end_of_utterance()
+        """
+        await self.send_message({"message": ClientMessageType.FORCE_END_OF_UTTERANCE})
+
     async def transcribe(
         self,
         source: BinaryIO,

sdk/rt/speechmatics/rt/_models.py

@@ -63,6 +63,10 @@ class ClientMessageType(str, Enum):
         SetRecognitionConfig: Updates transcription configuration during
             an active session (advanced use).
         GetSpeakers: Allows the client to request speaker data.
+        AddChannelAudio: Indicates that audio data follows (not used in message
+            headers, audio is sent as binary data).
+        EndOfChannel: Signals that no more audio data will be sent.
+        ForceEndOfUtterance: Signals that the end of an utterance has been reached.
 
     Examples:
         >>> # Starting a recognition session
@@ -86,6 +90,7 @@ class ClientMessageType(str, Enum):
     GET_SPEAKERS = "GetSpeakers"
     ADD_CHANNEL_AUDIO = "AddChannelAudio"
     END_OF_CHANNEL = "EndOfChannel"
+    FORCE_END_OF_UTTERANCE = "ForceEndOfUtterance"
 
 
 class ServerMessageType(str, Enum):