From 867837dcc41a42b43d08a18459a25853564d9c90 Mon Sep 17 00:00:00 2001
From: Storm <stormluykx@gmail.com>
Date: Tue, 30 Dec 2025 15:58:18 +0200
Subject: [PATCH] feat: implemented pause functionality in VAD agent

Functionality is implemented by pausing the _streaming_loop function.

ref: N25B-350
---
 .../agents/perception/vad_agent.py            | 40 +++++++++++++++++++
 src/control_backend/core/config.py            |  4 --
 2 files changed, 40 insertions(+), 4 deletions(-)

diff --git a/src/control_backend/agents/perception/vad_agent.py b/src/control_backend/agents/perception/vad_agent.py
index 8ccff0a..b2f1782 100644
--- a/src/control_backend/agents/perception/vad_agent.py
+++ b/src/control_backend/agents/perception/vad_agent.py
@@ -7,6 +7,7 @@ import zmq.asyncio as azmq
 
 from control_backend.agents import BaseAgent
 from control_backend.core.config import settings
+from control_backend.schemas.internal_message import InternalMessage
 
 from ...schemas.program_status import PROGRAM_STATUS, ProgramStatus
 from .transcription_agent.transcription_agent import TranscriptionAgent
@@ -86,6 +87,12 @@ class VADAgent(BaseAgent):
         self.audio_buffer = np.array([], dtype=np.float32)
         self.i_since_speech = settings.behaviour_settings.vad_initial_since_speech
         self._ready = asyncio.Event()
+
+        # Pause control
+        self._reset_needed = False
+        self._paused = asyncio.Event()
+        self._paused.set()  # Not paused at start
+
         self.model = None
 
     async def setup(self):
@@ -213,6 +220,16 @@ class VADAgent(BaseAgent):
         """
         await self._ready.wait()
         while self._running:
+            await self._paused.wait()
+
+            # After being unpaused, reset stream and buffers
+            if self._reset_needed:
+                self.logger.debug("Resuming: resetting stream and buffers.")
+                await self._reset_stream()
+                self.audio_buffer = np.array([], dtype=np.float32)
+                self.i_since_speech = settings.behaviour_settings.vad_initial_since_speech
+                self._reset_needed = False
+
             assert self.audio_in_poller is not None
             data = await self.audio_in_poller.poll()
             if data is None:
@@ -254,3 +271,26 @@ class VADAgent(BaseAgent):
             # At this point, we know that the speech has ended.
             # Prepend the last chunk that had no speech, for a more fluent boundary
             self.audio_buffer = chunk
+
+    async def handle_message(self, msg: InternalMessage):
+        """
+        Handle incoming messages.
+
+        Expects messages to pause or resume the VAD processing from User Interrupt Agent.
+
+        :param msg: The received internal message.
+        """
+        sender = msg.sender
+
+        if sender == settings.agent_settings.user_interrupt_name:
+            if msg.body == "PAUSE":
+                self.logger.info("Pausing VAD processing.")
+                self._paused.clear()
+                self._reset_needed = True
+            elif msg.body == "RESUME":
+                self.logger.info("Resuming VAD processing.")
+                self._paused.set()
+            else:
+                self.logger.warning(f"Unknown command from User Interrupt Agent: {msg.body}")
+        else:
+            self.logger.debug(f"Ignoring message from unknown sender: {sender}")
\ No newline at end of file
diff --git a/src/control_backend/core/config.py b/src/control_backend/core/config.py
index c8edfd7..8a7267c 100644
--- a/src/control_backend/core/config.py
+++ b/src/control_backend/core/config.py
@@ -100,14 +100,10 @@ class LLMSettings(BaseModel):
     """
 
     local_llm_url: str = "http://localhost:1234/v1/chat/completions"
-<<<<<<< HEAD
-    local_llm_model: str = "google/gemma-3-1b"
-=======
     local_llm_model: str = "gpt-oss"
     chat_temperature: float = 1.0
     code_temperature: float = 0.3
     n_parallel: int = 4
->>>>>>> feat/semantic-beliefs
 
 
 class VADSettings(BaseModel):