From b14741dca0ac6eeddf4b4561cff457ba552680bd Mon Sep 17 00:00:00 2001
From: vegu-ai-tools <152010387+vegu-ai-tools@users.noreply.github.com>
Date: Sun, 29 Mar 2026 13:38:29 +0300
Subject: [PATCH] Enhance TTS agent functionality by adding a context manager
 to suppress automatic voice generation during multi-turn planning in
 DirectorChatMixin. This prevents noisy voice synthesis and improves user
 experience. Also, introduced a property to check if auto generation is
 enabled.

---
 src/talemate/agents/director/chat/mixin.py | 29 ++++++++++++++++------
 src/talemate/agents/tts/__init__.py        | 27 ++++++++++++++++++++
 2 files changed, 48 insertions(+), 8 deletions(-)

diff --git a/src/talemate/agents/director/chat/mixin.py b/src/talemate/agents/director/chat/mixin.py
index 5b25e7d9..7b4c22a5 100644
--- a/src/talemate/agents/director/chat/mixin.py
+++ b/src/talemate/agents/director/chat/mixin.py
@@ -1,7 +1,9 @@
 import structlog
 from typing import Any, TYPE_CHECKING, Callable, Awaitable
 
+import talemate.instance as instance
 from talemate.agents.base import set_processing, AgentAction, AgentActionConfig
+from talemate.emit import emit
 from talemate.game.engine.nodes.core import GraphState
 
 from talemate.agents.director.action_core import utils as action_utils
@@ -957,12 +959,23 @@ class DirectorChatMixin:
         """
         Start arc generation. Like chat_generate_next but with @set_processing
         to ensure active_agent context is available.
+
+        Suppresses automatic TTS generation while arcs are being generated to
+        avoid noisy voice synthesis during long multi-step planning.
         """
-        return await self.chat_generate_next(
-            chat_id,
-            on_update=on_update,
-            on_done=on_done,
-            on_compacting=on_compacting,
-            on_compacted=on_compacted,
-            on_title_generated=on_title_generated,
-        )
+        tts_agent = instance.get_agent("tts")
+        if tts_agent.enabled and tts_agent.has_auto_generation:
+            emit(
+                "status",
+                "Automatic voice generation is paused during multi-turn generation.",
+                status="info",
+            )
+        with tts_agent.suppress_auto_generation():
+            return await self.chat_generate_next(
+                chat_id,
+                on_update=on_update,
+                on_done=on_done,
+                on_compacting=on_compacting,
+                on_compacted=on_compacted,
+                on_title_generated=on_title_generated,
+            )
diff --git a/src/talemate/agents/tts/__init__.py b/src/talemate/agents/tts/__init__.py
index 46bc929f..355051f4 100644
--- a/src/talemate/agents/tts/__init__.py
+++ b/src/talemate/agents/tts/__init__.py
@@ -2,6 +2,7 @@ from __future__ import annotations
 
 import asyncio
 import base64
+import contextlib
 import traceback
 import uuid
 from collections import deque
@@ -226,6 +227,7 @@ class TTSAgent(
         self.is_enabled = False  # tts agent is disabled by default
         self.actions = TTSAgent.init_actions()
         self.playback_done_event = asyncio.Event()
+        self._suppress_auto_generation: int = 0
 
         # Queue management for voice generation
         # Each queue instance gets a unique id so it can later be referenced
@@ -241,6 +243,19 @@ class TTSAgent(
         self._queue_task: asyncio.Task | None = None
         self._queue_lock = asyncio.Lock()
 
+    @contextlib.contextmanager
+    def suppress_auto_generation(self):
+        """Context manager to suppress automatic TTS generation.
+
+        While active, reactive generation triggered by game_loop_new_message
+        is skipped. Explicit calls to generate() still work.
+        """
+        self._suppress_auto_generation += 1
+        try:
+            yield
+        finally:
+            self._suppress_auto_generation -= 1
+
     # general helpers
 
     @property
@@ -283,6 +298,15 @@ class TTSAgent(
             self.actions["_config"].config["generate_for_context_investigation"].value
         )
 
+    @property
+    def has_auto_generation(self) -> bool:
+        return any([
+            self.generate_for_npc,
+            self.generate_for_player,
+            self.generate_for_narration,
+            self.generate_for_context_investigation,
+        ])
+
     @property
     def speaker_separation(self) -> str:
         return self.actions["_config"].config["speaker_separation"].value
@@ -468,6 +492,9 @@ class TTSAgent(
         Called when a conversation is generated
         """
 
+        if self._suppress_auto_generation:
+            return
+
         if self.scene.environment == "creative":
             return