Bump homekit-audio-proxy to 1.2.1

https://github.com/bdraco/homekit-audio-proxy/releases/tag/v1.2.1

homeassistant/components/homekit/manifest.json

@@ -11,6 +11,7 @@
   "requirements": [
     "HAP-python==5.0.0",
     "fnv-hash-fast==1.6.0",
+    "homekit-audio-proxy==1.2.1",
     "PyQRCode==1.2.1",
     "base36==0.1.1"
   ],

homeassistant/components/homekit/type_cameras.py

@@ -6,6 +6,7 @@ import logging
 from typing import Any
 
 from haffmpeg.core import FFMPEG_STDERR, HAFFmpeg
+from homekit_audio_proxy import AudioProxy
 from pyhap.camera import (
     VIDEO_CODEC_PARAM_LEVEL_TYPES,
     VIDEO_CODEC_PARAM_PROFILE_ID_TYPES,
@@ -89,11 +90,10 @@ AUDIO_OUTPUT = (
     "{a_application}"
     "-ac 1 -ar {a_sample_rate}k "
     "-b:a {a_max_bitrate}k -bufsize {a_bufsize}k "
+    "{a_frame_duration}"
     "-payload_type 110 "
     "-ssrc {a_ssrc} -f rtp "
-    "-srtp_out_suite AES_CM_128_HMAC_SHA1_80 -srtp_out_params {a_srtp_key} "
-    "srtp://{address}:{a_port}?rtcpport={a_port}&"
-    "localrtpport={a_port}&pkt_size={a_pkt_size}"
+    "rtp://127.0.0.1:{a_proxy_port}?pkt_size={a_pkt_size}"
 )
 
 SLOW_RESOLUTIONS = [
@@ -120,6 +120,7 @@ FFMPEG_WATCH_INTERVAL = timedelta(seconds=5)
 FFMPEG_LOGGER = "ffmpeg_logger"
 FFMPEG_WATCHER = "ffmpeg_watcher"
 FFMPEG_PID = "ffmpeg_pid"
+AUDIO_PROXY = "audio_proxy"
 SESSION_ID = "session_id"
 
 CONFIG_DEFAULTS = {
@@ -339,8 +340,33 @@ class Camera(HomeDoorbellAccessory, PyhapCamera):  # type: ignore[misc]
                 + " "
             )
         audio_application = ""
+        audio_frame_duration = ""
         if self.config[CONF_AUDIO_CODEC] == "libopus":
             audio_application = "-application lowdelay "
+            audio_frame_duration = (
+                f"-frame_duration {stream_config.get('a_packet_time', 20)} "
+            )
+        # Start audio proxy to convert Opus RTP timestamps from 48kHz
+        # (FFmpeg's hardcoded Opus RTP clock rate per RFC 7587) to the
+        # sample rate negotiated by HomeKit (typically 16kHz).
+        # a_sample_rate is in kHz (e.g. 16 for 16000 Hz) from pyhap TLV.
+        audio_proxy: AudioProxy | None = None
+        if self.config[CONF_SUPPORT_AUDIO]:
+            audio_proxy = AudioProxy(
+                dest_addr=stream_config["address"],
+                dest_port=stream_config["a_port"],
+                srtp_key_b64=stream_config["a_srtp_key"],
+                target_clock_rate=stream_config["a_sample_rate"] * 1000,
+            )
+            await audio_proxy.async_start()
+            if not audio_proxy.local_port:
+                _LOGGER.error(
+                    "[%s] Audio proxy failed to start",
+                    self.display_name,
+                )
+                await audio_proxy.async_stop()
+                audio_proxy = None
+
         output_vars = stream_config.copy()
         output_vars.update(
             {
@@ -354,6 +380,8 @@ class Camera(HomeDoorbellAccessory, PyhapCamera):  # type: ignore[misc]
                 "a_pkt_size": self.config[CONF_AUDIO_PACKET_SIZE],
                 "a_encoder": self.config[CONF_AUDIO_CODEC],
                 "a_application": audio_application,
+                "a_frame_duration": audio_frame_duration,
+                "a_proxy_port": audio_proxy.local_port if audio_proxy else 0,
             }
         )
         output = VIDEO_OUTPUT.format(**output_vars)
@@ -371,6 +399,8 @@ class Camera(HomeDoorbellAccessory, PyhapCamera):  # type: ignore[misc]
         )
         if not opened:
             _LOGGER.error("Failed to open ffmpeg stream")
+            if audio_proxy:
+                await audio_proxy.async_stop()
             return False
 
         _LOGGER.debug(
@@ -381,6 +411,7 @@ class Camera(HomeDoorbellAccessory, PyhapCamera):  # type: ignore[misc]
 
         session_info["stream"] = stream
         session_info[FFMPEG_PID] = stream.process.pid
+        session_info[AUDIO_PROXY] = audio_proxy
 
         stderr_reader = await stream.get_reader(source=FFMPEG_STDERR)
 
@@ -441,6 +472,9 @@ class Camera(HomeDoorbellAccessory, PyhapCamera):  # type: ignore[misc]
     async def stop_stream(self, session_info: dict[str, Any]) -> None:
         """Stop the stream for the given ``session_id``."""
         session_id = session_info["id"]
+        if proxy := session_info.pop(AUDIO_PROXY, None):
+            await proxy.async_stop()
+
         if not (stream := session_info.get("stream")):
             _LOGGER.debug("No stream for session ID %s", session_id)
             return

requirements_all.txt

@@ -1228,6 +1228,9 @@ home-assistant-frontend==20260304.0
 # homeassistant.components.conversation
 home-assistant-intents==2026.3.3
 
+# homeassistant.components.homekit
+homekit-audio-proxy==1.2.1
+
 # homeassistant.components.gentex_homelink
 homelink-integration-api==0.0.1
 

requirements_test_all.txt

@@ -1089,6 +1089,9 @@ home-assistant-frontend==20260304.0
 # homeassistant.components.conversation
 home-assistant-intents==2026.3.3
 
+# homeassistant.components.homekit
+homekit-audio-proxy==1.2.1
+
 # homeassistant.components.gentex_homelink
 homelink-integration-api==0.0.1
 

tests/components/homekit/test_type_cameras.py

@@ -46,6 +46,7 @@ from homeassistant.util import dt as dt_util
 
 from tests.components.camera.common import mock_turbo_jpeg
 
+MOCK_AUDIO_PROXY_PORT = 23456
 MOCK_START_STREAM_TLV = "ARUCAQEBEDMD1QMXzEaatnKSQ2pxovYCNAEBAAIJAQECAgECAwEAAwsBAgAFAgLQAgMBHgQXAQFjAgQ768/RAwIrAQQEAAAAPwUCYgUDLAEBAwIMAQEBAgEAAwECBAEUAxYBAW4CBCzq28sDAhgABAQAAKBABgENBAEA"
 MOCK_END_POINTS_TLV = "ARAzA9UDF8xGmrZykkNqcaL2AgEAAxoBAQACDTE5Mi4xNjguMjA4LjUDAi7IBAKkxwQlAQEAAhDN0+Y0tZ4jzoO0ske9UsjpAw6D76oVXnoi7DbawIG4CwUlAQEAAhCyGcROB8P7vFRDzNF2xrK1Aw6NdcLugju9yCfkWVSaVAYEDoAsAAcEpxV8AA=="
 MOCK_START_STREAM_SESSION_UUID = UUID("3303d503-17cc-469a-b672-92436a71a2f6")
@@ -59,6 +60,21 @@ async def setup_homeassistant(hass: HomeAssistant) -> None:
     await async_setup_component(hass, "homeassistant", {})
 
 
+@pytest.fixture(autouse=True)
+def mock_audio_proxy():
+    """Mock AudioProxy to avoid spawning real subprocesses."""
+    mock_proxy = MagicMock()
+    mock_proxy.local_port = MOCK_AUDIO_PROXY_PORT
+    mock_proxy.async_start = AsyncMock()
+    mock_proxy.async_stop = AsyncMock()
+
+    with patch(
+        "homeassistant.components.homekit.type_cameras.AudioProxy",
+        return_value=mock_proxy,
+    ):
+        yield mock_proxy
+
+
 async def _async_start_streaming(hass: HomeAssistant, acc: Camera) -> None:
     """Start streaming a camera."""
     acc.set_selected_stream_configuration(MOCK_START_STREAM_TLV)
@@ -206,10 +222,9 @@ async def test_camera_stream_source_configured(hass: HomeAssistant, run_driver)
         "rtp -srtp_out_suite AES_CM_128_HMAC_SHA1_80 -srtp_out_params "
         "zdPmNLWeI86DtLJHvVLI6YPvqhVeeiLsNtrAgbgL "
         "srtp://192.168.208.5:51246?rtcpport=51246&localrtpport=51246&pkt_size=1316 -map 0:a:0 "
-        "-vn -c:a libopus -application lowdelay -ac 1 -ar 24k -b:a 24k -bufsize 96k -payload_type "
-        "110 -ssrc {a_ssrc} -f rtp -srtp_out_suite AES_CM_128_HMAC_SHA1_80 -srtp_out_params "
-        "shnETgfD+7xUQ8zRdsaytY11wu6CO73IJ+RZVJpU "
-        "srtp://192.168.208.5:51108?rtcpport=51108&localrtpport=51108&pkt_size=188"
+        "-vn -c:a libopus -application lowdelay -ac 1 -ar 24k -b:a 24k -bufsize 96k "
+        f"-frame_duration 20 -payload_type 110 -ssrc {{a_ssrc}} -f rtp "
+        f"rtp://127.0.0.1:{MOCK_AUDIO_PROXY_PORT}?pkt_size=188"
     )
 
     working_ffmpeg.open.assert_called_with(
@@ -322,6 +337,52 @@ async def test_camera_stream_source_configured_with_failing_ffmpeg(
         await _async_stop_all_streams(hass, acc)
 
 
+async def test_camera_stream_source_audio_proxy_fails(
+    hass: HomeAssistant, run_driver, mock_audio_proxy: MagicMock
+) -> None:
+    """Test streaming continues without audio when audio proxy fails to start."""
+    await async_setup_component(hass, ffmpeg.DOMAIN, {ffmpeg.DOMAIN: {}})
+    await async_setup_component(
+        hass, camera.DOMAIN, {camera.DOMAIN: {"platform": "demo"}}
+    )
+    await hass.async_block_till_done()
+
+    entity_id = "camera.demo_camera"
+
+    hass.states.async_set(entity_id, None)
+    await hass.async_block_till_done()
+    acc = Camera(
+        hass,
+        run_driver,
+        "Camera",
+        entity_id,
+        2,
+        {CONF_STREAM_SOURCE: "/dev/null", CONF_SUPPORT_AUDIO: True},
+    )
+
+    acc.run()
+    await _async_setup_endpoints(hass, acc)
+
+    # Simulate audio proxy failing to bind a port
+    mock_audio_proxy.local_port = 0
+    working_ffmpeg = _get_working_mock_ffmpeg()
+
+    with (
+        patch(
+            "homeassistant.components.demo.camera.DemoCamera.stream_source",
+            return_value=None,
+        ),
+        patch(
+            "homeassistant.components.homekit.type_cameras.HAFFmpeg",
+            return_value=working_ffmpeg,
+        ),
+    ):
+        await _async_start_streaming(hass, acc)
+        await _async_stop_all_streams(hass, acc)
+
+    mock_audio_proxy.async_stop.assert_called()
+
+
 async def test_camera_stream_source_found(hass: HomeAssistant, run_driver) -> None:
     """Test a camera that can stream and we get the source from the entity."""
     await async_setup_component(hass, ffmpeg.DOMAIN, {ffmpeg.DOMAIN: {}})
@@ -540,10 +601,9 @@ async def test_camera_stream_source_configured_and_copy_codec(
         "-bufsize 1196k -maxrate 299k -payload_type 99 -ssrc {v_ssrc} -f rtp -srtp_out_suite "
         "AES_CM_128_HMAC_SHA1_80 -srtp_out_params zdPmNLWeI86DtLJHvVLI6YPvqhVeeiLsNtrAgbgL "
         "srtp://192.168.208.5:51246?rtcpport=51246&localrtpport=51246&pkt_size=1316 -map 0:a:0 "
-        "-vn -c:a copy -ac 1 -ar 24k -b:a 24k -bufsize 96k -payload_type 110 -ssrc {a_ssrc} "
-        "-f rtp -srtp_out_suite AES_CM_128_HMAC_SHA1_80 -srtp_out_params "
-        "shnETgfD+7xUQ8zRdsaytY11wu6CO73IJ+RZVJpU "
-        "srtp://192.168.208.5:51108?rtcpport=51108&localrtpport=51108&pkt_size=188"
+        "-vn -c:a copy -ac 1 -ar 24k -b:a 24k -bufsize 96k "
+        f"-payload_type 110 -ssrc {{a_ssrc}} -f rtp "
+        f"rtp://127.0.0.1:{MOCK_AUDIO_PROXY_PORT}?pkt_size=188"
     )
 
     working_ffmpeg.open.assert_called_with(
@@ -616,10 +676,9 @@ async def test_camera_stream_source_configured_and_override_profile_names(
         "-bufsize 1196k -maxrate 299k -payload_type 99 -ssrc {v_ssrc} -f rtp -srtp_out_suite "
         "AES_CM_128_HMAC_SHA1_80 -srtp_out_params zdPmNLWeI86DtLJHvVLI6YPvqhVeeiLsNtrAgbgL "
         "srtp://192.168.208.5:51246?rtcpport=51246&localrtpport=51246&pkt_size=1316 -map 0:a:0 "
-        "-vn -c:a copy -ac 1 -ar 24k -b:a 24k -bufsize 96k -payload_type 110 -ssrc {a_ssrc} "
-        "-f rtp -srtp_out_suite AES_CM_128_HMAC_SHA1_80 -srtp_out_params "
-        "shnETgfD+7xUQ8zRdsaytY11wu6CO73IJ+RZVJpU "
-        "srtp://192.168.208.5:51108?rtcpport=51108&localrtpport=51108&pkt_size=188"
+        "-vn -c:a copy -ac 1 -ar 24k -b:a 24k -bufsize 96k "
+        f"-payload_type 110 -ssrc {{a_ssrc}} -f rtp "
+        f"rtp://127.0.0.1:{MOCK_AUDIO_PROXY_PORT}?pkt_size=188"
     )
 
     working_ffmpeg.open.assert_called_with(
@@ -693,10 +752,9 @@ async def test_camera_streaming_fails_after_starting_ffmpeg(
         "-bufsize 1196k -maxrate 299k -payload_type 99 -ssrc {v_ssrc} -f rtp -srtp_out_suite "
         "AES_CM_128_HMAC_SHA1_80 -srtp_out_params zdPmNLWeI86DtLJHvVLI6YPvqhVeeiLsNtrAgbgL "
         "srtp://192.168.208.5:51246?rtcpport=51246&localrtpport=51246&pkt_size=1316 -map 0:a:0 "
-        "-vn -c:a copy -ac 1 -ar 24k -b:a 24k -bufsize 96k -payload_type 110 -ssrc {a_ssrc} "
-        "-f rtp -srtp_out_suite AES_CM_128_HMAC_SHA1_80 -srtp_out_params "
-        "shnETgfD+7xUQ8zRdsaytY11wu6CO73IJ+RZVJpU "
-        "srtp://192.168.208.5:51108?rtcpport=51108&localrtpport=51108&pkt_size=188"
+        "-vn -c:a copy -ac 1 -ar 24k -b:a 24k -bufsize 96k "
+        f"-payload_type 110 -ssrc {{a_ssrc}} -f rtp "
+        f"rtp://127.0.0.1:{MOCK_AUDIO_PROXY_PORT}?pkt_size=188"
     )
 
     ffmpeg_with_invalid_pid.open.assert_called_with(