dv-pipecat-ai 0.0.85.dev11__py3-none-any.whl → 0.0.85.dev12__py3-none-any.whl

{dv_pipecat_ai-0.0.85.dev11.dist-info → dv_pipecat_ai-0.0.85.dev12.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dv-pipecat-ai
-Version: 0.0.85.dev11
+Version: 0.0.85.dev12
 Summary: An open source framework for voice (and multimodal) assistants
 License-Expression: BSD-2-Clause
 Project-URL: Source, https://github.com/pipecat-ai/pipecat

{dv_pipecat_ai-0.0.85.dev11.dist-info → dv_pipecat_ai-0.0.85.dev12.dist-info}/RECORD

@@ -1,4 +1,4 @@
-dv_pipecat_ai-0.0.85.dev11.dist-info/licenses/LICENSE,sha256=DWY2QGf2eMCFhuu2ChairtT6CB7BEFffNVhXWc4Od08,1301
+dv_pipecat_ai-0.0.85.dev12.dist-info/licenses/LICENSE,sha256=DWY2QGf2eMCFhuu2ChairtT6CB7BEFffNVhXWc4Od08,1301
 pipecat/__init__.py,sha256=j0Xm6adxHhd7D06dIyyPV_GlBYLlBnTAERVvD_jAARQ,861
 pipecat/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 pipecat/adapters/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -77,7 +77,7 @@ pipecat/extensions/voicemail/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NM
 pipecat/extensions/voicemail/voicemail_detector.py,sha256=g3L1m3cPJzsadeB5a8WRC9klH0D8m7xfPgB2YEaL6Do,29983
 pipecat/frames/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 pipecat/frames/frames.proto,sha256=JXZm3VXLR8zMOUcOuhVoe2mhM3MQIQGMJXLopdJO_5Y,839
-pipecat/frames/frames.py,sha256=oqoo7p-uJOqak50mxhCGq7S0TusM0I4qp3QAftKHQnw,45428
+pipecat/frames/frames.py,sha256=2aXsBpZB6dU7I1PZRFh8RqALDktxxoRgZPa1pf_cdCM,45804
 pipecat/frames/protobufs/frames_pb2.py,sha256=VHgGV_W7qQ4sfQK6RHb5_DggLm3PiSYMr6aBZ8_p1cQ,2590
 pipecat/metrics/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 pipecat/metrics/metrics.py,sha256=bdZNciEtLTtA-xgoKDz2RJAy6fKrXkTwz3pryVHzc2M,2713
@@ -111,7 +111,7 @@ pipecat/processors/idle_frame_processor.py,sha256=z8AuhGap61lA5K35P6XCaOpn4kkmK_
 pipecat/processors/logger.py,sha256=VGNwxQSc_F0rS3KBmfqas7f5aFyRQKfeljozOxfGXk4,2393
 pipecat/processors/producer_processor.py,sha256=iIIOHZd77APvUGP7JqFbznAHUnCULcq_qYiSEjwXHcc,3265
 pipecat/processors/text_transformer.py,sha256=LnfWJYzntJhZhrQ1lgSSY4D4VbHtrQJgrC227M69ZYU,1718
-pipecat/processors/transcript_processor.py,sha256=CG9yej6WOiy_HhagNXjxkISHkHii0JDfK_V6opseC2E,11740
+pipecat/processors/transcript_processor.py,sha256=SDbqFLzasptZWqeiA6BESJ4hky-Uh-ZUNYBZR0q0Dnw,12508
 pipecat/processors/two_stage_user_idle_processor.py,sha256=uf2aZh_lfW-eMxmFogP3R4taAJ1yXOSqjKsR7oXtD0Y,2938
 pipecat/processors/user_idle_processor.py,sha256=PQItBx5bL1y_lFTtHf0zgLubjCXv60jU1jrVtm-K4bg,9268
 pipecat/processors/aggregators/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -119,7 +119,7 @@ pipecat/processors/aggregators/dtmf_aggregator.py,sha256=nngjLiaOtcZtuCNpYPyfUVL
 pipecat/processors/aggregators/gated.py,sha256=tii0sRrBkRW6y9Xq5iTWPnqlOEejU4VqPIPtdOa61pc,3073
 pipecat/processors/aggregators/gated_openai_llm_context.py,sha256=cr6MT8J6SpPzZbppKPOKe3_pt_5qXC9g6a4wvZDyrec,3005
 pipecat/processors/aggregators/llm_context.py,sha256=eDf1cQElcISLx3onaA9LCWuepzb2G_JGszLzpNXggXo,9723
-pipecat/processors/aggregators/llm_response.py,sha256=0StzYtq7EzlAFSWp10I0yY0pV1jysw1ySEWv5R50h_s,47360
+pipecat/processors/aggregators/llm_response.py,sha256=W0bqc5IZTE1cIB8egBefUOaPZiLklzge5npFWvAbCcw,48248
 pipecat/processors/aggregators/llm_response_universal.py,sha256=fBnB3rZVdxj4iEKIWcnR7yTpqyKupbcg7IUv6XVxrDQ,34287
 pipecat/processors/aggregators/openai_llm_context.py,sha256=cC8DXdVPERRN04i0i-1Ys6kusvnbMALeH-Z8Pu5K684,12999
 pipecat/processors/aggregators/sentence.py,sha256=E7e3knfQl6HEGpYMKPklF1aO_gOn-rr7SnynErwfkQk,2235
@@ -384,7 +384,7 @@ pipecat/utils/tracing/service_decorators.py,sha256=HwDCqLGijhYD3F8nxDuQmEw-YkRw0
 pipecat/utils/tracing/setup.py,sha256=7TEgPNpq6M8lww8OQvf0P9FzYc5A30xICGklVA-fua0,2892
 pipecat/utils/tracing/turn_context_provider.py,sha256=ikon3plFOx0XbMrH6DdeHttNpb-U0gzMZIm3bWLc9eI,2485
 pipecat/utils/tracing/turn_trace_observer.py,sha256=dma16SBJpYSOE58YDWy89QzHyQFc_9gQZszKeWixuwc,9725
-dv_pipecat_ai-0.0.85.dev11.dist-info/METADATA,sha256=_scIy5gP8k7GUtLAA9NzNVT_T1y__8ROU0gPj1G6FCw,32858
-dv_pipecat_ai-0.0.85.dev11.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-dv_pipecat_ai-0.0.85.dev11.dist-info/top_level.txt,sha256=kQzG20CxGf-nSsHmtXHx3hY2-8zHA3jYg8jk0TajqXc,8
-dv_pipecat_ai-0.0.85.dev11.dist-info/RECORD,,
+dv_pipecat_ai-0.0.85.dev12.dist-info/METADATA,sha256=pj9DcBZS3A1SFUXm5aZDJS5K9e4YldFoPBcrrBiFxPI,32858
+dv_pipecat_ai-0.0.85.dev12.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+dv_pipecat_ai-0.0.85.dev12.dist-info/top_level.txt,sha256=kQzG20CxGf-nSsHmtXHx3hY2-8zHA3jYg8jk0TajqXc,8
+dv_pipecat_ai-0.0.85.dev12.dist-info/RECORD,,

pipecat/frames/frames.py

@@ -451,12 +451,14 @@ class TranscriptionMessage:
         content: The message content/text.
         user_id: Optional identifier for the user.
         timestamp: Optional timestamp when the message was created.
+        message_id: Optional unique identifier for tracking and dropping messages.
     """
 
     role: Literal["user", "assistant"]
     content: str
     user_id: Optional[str] = None
     timestamp: Optional[str] = None
+    message_id: Optional[int] = None
 
 
 @dataclass
@@ -510,6 +512,17 @@ class TranscriptionUpdateFrame(DataFrame):
         return f"{self.name}(pts: {pts}, messages: {len(self.messages)})"
 
 
+@dataclass
+class TranscriptDropFrame(DataFrame):
+    """Frame indicating previously emitted transcript chunks should be discarded.
+
+    Parameters:
+        transcript_ids: List of frame/message identifiers to drop.
+    """
+
+    transcript_ids: List[int]
+
+
 @dataclass
 class LLMContextFrame(Frame):
     """Frame containing a universal LLM context.

pipecat/processors/aggregators/llm_response.py

@@ -51,6 +51,7 @@ from pipecat.frames.frames import (
     StartFrame,
     StartInterruptionFrame,
     TextFrame,
+    TranscriptDropFrame,
     TranscriptionFrame,
     UserImageRawFrame,
     UserStartedSpeakingFrame,
@@ -446,6 +447,7 @@ class LLMUserContextAggregator(LLMContextResponseAggregator):
         self._latest_final_transcript = ""
         self._last_user_speaking_time = 0
         self._last_aggregation_push_time = 0
+        self._pending_transcription_ids: List[int] = []
 
     async def reset(self):
         """Reset the aggregation state and interruption strategies."""
@@ -453,6 +455,7 @@ class LLMUserContextAggregator(LLMContextResponseAggregator):
         self._was_bot_speaking = False
         self._seen_interim_results = False
         self._waiting_for_aggregation = False
+        self._pending_transcription_ids.clear()
         [await s.reset() for s in self._interruption_strategies]
 
     async def handle_aggregation(self, aggregation: str):
@@ -548,7 +551,8 @@ class LLMUserContextAggregator(LLMContextResponseAggregator):
                     await self._process_aggregation()
                 else:
                     self.logger.debug("Interruption conditions not met - not pushing aggregation")
-                    # Don't process aggregation, just reset it
+                    # Don't process aggregation, discard pending transcriptions and reset
+                    await self._discard_pending_transcriptions("interruption_conditions_not_met")
                     await self.reset()
             else:
                 if trigger_interruption:
@@ -614,10 +618,18 @@ class LLMUserContextAggregator(LLMContextResponseAggregator):
         for s in self.interruption_strategies:
             await s.append_audio(frame.audio, frame.sample_rate)
 
+    async def _discard_pending_transcriptions(self, reason: str):
+        """Notify upstream processors that pending transcripts should be dropped."""
+        if self._pending_transcription_ids:
+            drop_frame = TranscriptDropFrame(transcript_ids=list(self._pending_transcription_ids))
+            await self.push_frame(drop_frame, FrameDirection.UPSTREAM)
+        self._pending_transcription_ids.clear()
+
     async def _handle_user_started_speaking(self, frame: UserStartedSpeakingFrame):
         if len(self._aggregation) > 0:
             self.logger.debug(f"Dropping {self._aggregation}")
             self._aggregation = ""
+            await self._discard_pending_transcriptions("user_started_speaking")
         self._latest_final_transcript = ""
         self._last_user_speaking_time = time.time()
         self._user_speaking = True
@@ -662,6 +674,7 @@ class LLMUserContextAggregator(LLMContextResponseAggregator):
             return
 
         self._aggregation += f" {text}" if self._aggregation else text
+        self._pending_transcription_ids.append(frame.id)
         # We just got a final result, so let's reset interim results.
         self._seen_interim_results = False
 
@@ -791,6 +804,7 @@ class LLMUserContextAggregator(LLMContextResponseAggregator):
             if self._bot_speaking and not self._params.enable_emulated_vad_interruptions:
                 # If emulated VAD interruptions are disabled and bot is speaking, ignore
                 logger.debug("Ignoring user speaking emulation, bot is speaking.")
+                await self._discard_pending_transcriptions("emulated_vad_ignored")
                 await self.reset()
             else:
                 # Either bot is not speaking, or emulated VAD interruptions are enabled

pipecat/processors/transcript_processor.py

@@ -20,6 +20,7 @@ from pipecat.frames.frames import (
     EndFrame,
     Frame,
     StartInterruptionFrame,
+    TranscriptDropFrame,
     TranscriptionFrame,
     TranscriptionMessage,
     TranscriptionUpdateFrame,
@@ -44,6 +45,7 @@ class BaseTranscriptProcessor(FrameProcessor):
         super().__init__(**kwargs)
         self._processed_messages: List[TranscriptionMessage] = []
         self._register_event_handler("on_transcript_update")
+        self._register_event_handler("on_transcript_drop")
 
     async def _emit_update(self, messages: List[TranscriptionMessage]):
         """Emit transcript updates for new messages.
@@ -57,6 +59,19 @@ class BaseTranscriptProcessor(FrameProcessor):
             await self._call_event_handler("on_transcript_update", update_frame)
             await self.push_frame(update_frame)
 
+    async def _handle_transcript_drop(self, frame: TranscriptDropFrame):
+        """Handle transcript drop notifications by removing stored messages."""
+        if not frame.transcript_ids:
+            return
+
+        await self._call_event_handler("on_transcript_drop", frame)
+
+        drop_ids = set(frame.transcript_ids)
+        if drop_ids:
+            self._processed_messages = [
+                msg for msg in self._processed_messages if msg.message_id not in drop_ids
+            ]
+
 
 class UserTranscriptProcessor(BaseTranscriptProcessor):
     """Processes user transcription frames into timestamped conversation messages."""
@@ -72,9 +87,15 @@ class UserTranscriptProcessor(BaseTranscriptProcessor):
 
         if isinstance(frame, TranscriptionFrame):
             message = TranscriptionMessage(
-                role="user", user_id=frame.user_id, content=frame.text, timestamp=frame.timestamp
+                role="user",
+                user_id=frame.user_id,
+                content=frame.text,
+                timestamp=frame.timestamp,
+                message_id=frame.id,
             )
             await self._emit_update([message])
+        elif isinstance(frame, TranscriptDropFrame):
+            await self._handle_transcript_drop(frame)
 
         await self.push_frame(frame, direction)