diff --git a/gateway/run.py b/gateway/run.py
index cef8dc845fa6..9d75774010f1 100644
--- a/gateway/run.py
+++ b/gateway/run.py
@@ -17968,7 +17968,17 @@ class GatewayRunner(GatewayAuthorizationMixin, GatewayKanbanWatchersMixin, Gatew
                 return f"{prefix}\n\n{user_text}", []
             return prefix, []
 
-        from tools.transcription_tools import transcribe_audio
+        try:
+            from tools.transcription_tools import transcribe_audio
+        except ModuleNotFoundError as e:
+            logger.error("Transcription module unavailable: %s", e)
+            unavailable_note = "[voice message could not be transcribed]"
+            _placeholder = "(The user sent a message with no text content)"
+            if user_text and user_text.strip() == _placeholder:
+                return unavailable_note, []
+            if user_text:
+                return f"{unavailable_note}\n\n{user_text}", []
+            return unavailable_note, []
 
         enriched_parts = []
         successful_transcripts: List[str] = []
diff --git a/tests/gateway/test_stt_config.py b/tests/gateway/test_stt_config.py
index 5006eafee319..3a0b54ea69a6 100644
--- a/tests/gateway/test_stt_config.py
+++ b/tests/gateway/test_stt_config.py
@@ -143,6 +143,44 @@ async def test_enrich_message_with_transcription_returns_tuple_for_empty_content
     assert transcripts == ["hello from a captionless voice note"]
 
 
+@pytest.mark.parametrize(
+    ("user_text", "expected_text"),
+    [
+        ("caption", "[voice message could not be transcribed]\n\ncaption"),
+        ("", "[voice message could not be transcribed]"),
+        (
+            "(The user sent a message with no text content)",
+            "[voice message could not be transcribed]",
+        ),
+    ],
+)
+@pytest.mark.asyncio
+async def test_enrich_message_with_transcription_handles_missing_transcription_module_gracefully(
+    user_text,
+    expected_text,
+):
+    from gateway.run import GatewayRunner
+
+    runner = GatewayRunner.__new__(GatewayRunner)
+    runner.config = GatewayConfig(stt_enabled=True)
+
+    real_import = __import__
+
+    def fake_import(name, globals=None, locals=None, fromlist=(), level=0):
+        if name == "tools.transcription_tools":
+            raise ModuleNotFoundError("No module named 'tools.transcription_tools'")
+        return real_import(name, globals, locals, fromlist, level)
+
+    with patch("builtins.__import__", side_effect=fake_import):
+        result, transcripts = await runner._enrich_message_with_transcription(
+            user_text,
+            ["/tmp/voice.ogg"],
+        )
+
+    assert result == expected_text
+    assert transcripts == []
+
+
 @pytest.mark.asyncio
 async def test_prepare_inbound_message_text_transcribes_queued_voice_event():
     from gateway.run import GatewayRunner