diff --git a/src/tuimble/audio.py b/src/tuimble/audio.py
index 7930d78..1715d71 100644
--- a/src/tuimble/audio.py
+++ b/src/tuimble/audio.py
@@ -39,6 +39,7 @@ class AudioPipeline:
         self._input_stream = None
         self._output_stream = None
         self._capturing = False
+        self._deafened = False
 
     def start(self):
         """Open audio streams."""
@@ -84,6 +85,14 @@ class AudioPipeline:
     def capturing(self, value: bool):
         self._capturing = value
 
+    @property
+    def deafened(self) -> bool:
+        return self._deafened
+
+    @deafened.setter
+    def deafened(self, value: bool):
+        self._deafened = value
+
     def _capture_callback(self, indata, frames, time_info, status):
         """Called by sounddevice when input data is available."""
         if status:
@@ -98,6 +107,9 @@ class AudioPipeline:
         """Called by sounddevice when output buffer needs data."""
         if status:
             log.warning("playback status: %s", status)
+        if self._deafened:
+            outdata[:] = b"\x00" * len(outdata)
+            return
         try:
             pcm = self._playback_queue.get_nowait()
             n = min(len(pcm), len(outdata))
@@ -116,6 +128,8 @@ class AudioPipeline:
 
     def queue_playback(self, pcm_data: bytes):
         """Queue raw PCM data for playback (16-bit, mono, 48kHz)."""
+        if self._deafened:
+            return
         try:
             self._playback_queue.put_nowait(pcm_data)
         except queue.Full:
diff --git a/tests/test_audio.py b/tests/test_audio.py
index 3d6db05..b93d359 100644
--- a/tests/test_audio.py
+++ b/tests/test_audio.py
@@ -10,6 +10,7 @@ def test_default_construction():
     assert ap._input_device is None
     assert ap._output_device is None
     assert ap.capturing is False
+    assert ap.deafened is False
 
 
 def test_custom_construction():
@@ -89,6 +90,37 @@ def test_queue_playback_overflow_drops():
     assert ap._playback_queue.qsize() == ap._playback_queue.maxsize
 
 
+def test_deafened_toggle():
+    ap = AudioPipeline()
+    assert ap.deafened is False
+    ap.deafened = True
+    assert ap.deafened is True
+    ap.deafened = False
+    assert ap.deafened is False
+
+
+def test_queue_playback_discards_when_deafened():
+    """Incoming PCM is dropped when deafened."""
+    ap = AudioPipeline()
+    ap.deafened = True
+    ap.queue_playback(b"\x42" * 100)
+    assert ap._playback_queue.qsize() == 0
+
+
+def test_playback_callback_silence_when_deafened():
+    """Playback callback writes silence when deafened, even with queued data."""
+    ap = AudioPipeline()
+    frame_bytes = FRAME_SIZE * 2
+    # Queue data before deafening
+    pcm = b"\x42" * frame_bytes
+    ap.queue_playback(pcm)
+    ap.deafened = True
+
+    outdata = bytearray(b"\xff" * frame_bytes)
+    ap._playback_callback(outdata, FRAME_SIZE, None, None)
+    assert outdata == bytearray(frame_bytes)  # all zeros
+
+
 def test_stop_without_start():
     """Stop on unstarted pipeline should not raise."""
     ap = AudioPipeline()