Fix high memory and cpu usage from analyze jobs (#1157)

diff --git a/music_assistant/server/helpers/audio.py b/music_assistant/server/helpers/audio.py
index cb2c9857c0c186b4c7bc9db004845b6fc48ebc7c..2e9062e977eb8b0743309abc9d2cba92bf8e2e48 100644 (file)
--- a/music_assistant/server/helpers/audio.py
+++ b/music_assistant/server/helpers/audio.py
@@ -50,7 +50,6 @@ if TYPE_CHECKING:
     from music_assistant.server import MusicAssistant
 
 LOGGER = logging.getLogger(f"{ROOT_LOGGER_NAME}.audio")
-analyze_jobs: set[str] = set()
 # pylint:disable=consider-using-f-string,too-many-locals,too-many-statements
 
 
@@ -180,60 +179,6 @@ async def strip_silence(
     return stripped_data
 
 
-async def analyze_loudness(mass: MusicAssistant, streamdetails: StreamDetails) -> None:
-    """Analyze track audio to calculate EBU R128 loudness."""
-    if streamdetails.uri in analyze_jobs:
-        return
-    if len(analyze_jobs) >= 5:
-        LOGGER.debug("Skip analyzing EBU R128 loudness: max number of jobs reached")
-        return
-    try:
-        analyze_jobs.add(streamdetails.uri)
-        item_name = f"{streamdetails.provider}/{streamdetails.item_id}"
-        LOGGER.debug("Start analyzing EBU R128 loudness for %s", item_name)
-        # calculate EBU R128 integrated loudness with ffmpeg
-        ffmpeg_args = get_ffmpeg_args(
-            input_format=streamdetails.audio_format,
-            output_format=streamdetails.audio_format,
-            filter_params=["loudnorm=print_format=json"],
-            extra_args=["-t", "600"],  # limit to 10 minutes to prevent OOM
-            input_path=streamdetails.direct or "-",
-            output_path="NULL",
-        )
-        async with AsyncProcess(
-            ffmpeg_args,
-            enable_stdin=streamdetails.direct is None,
-            enable_stdout=False,
-            enable_stderr=True,
-        ) as ffmpeg_proc:
-            if streamdetails.direct is None:
-                music_prov = mass.get_provider(streamdetails.provider)
-                chunk_count = 0
-                async for audio_chunk in music_prov.get_audio_stream(streamdetails):
-                    chunk_count += 1
-                    await ffmpeg_proc.write(audio_chunk)
-                    if chunk_count == 600:
-                        # safety guard: max (more or less) 10 minutes of audio may be analyzed!
-                        break
-                await ffmpeg_proc.write_eof()
-
-            _, stderr = await ffmpeg_proc.communicate()
-            if loudness_details := _parse_loudnorm(stderr):
-                LOGGER.debug("Loudness measurement for %s: %s", item_name, loudness_details)
-                streamdetails.loudness = loudness_details
-                await mass.music.set_track_loudness(
-                    streamdetails.item_id, streamdetails.provider, loudness_details
-                )
-            else:
-                LOGGER.warning(
-                    "Could not determine EBU R128 loudness of %s - %s",
-                    item_name,
-                    stderr.decode() or "received empty value",
-                )
-    finally:
-        analyze_jobs.discard(streamdetails.uri)
-
-
 async def get_stream_details(
     mass: MusicAssistant,
     queue_item: QueueItem,
@@ -531,8 +476,8 @@ async def get_media_stream(  # noqa: PLR0915
             finished = False
         elif loudness_details := _parse_loudnorm(stderr):
             logger.log(VERBOSE_LOG_LEVEL, stderr.decode())
-            required_seconds = 300 if streamdetails.media_type == MediaType.RADIO else 60
-            if finished or seconds_streamed >= required_seconds:
+            required_seconds = 600 if streamdetails.media_type == MediaType.RADIO else 120
+            if finished or (seconds_streamed >= required_seconds):
                 LOGGER.debug("Loudness measurement for %s: %s", streamdetails.uri, loudness_details)
                 streamdetails.loudness = loudness_details
                 await mass.music.set_track_loudness(
@@ -551,11 +496,6 @@ async def get_media_stream(  # noqa: PLR0915
             if music_prov := mass.get_provider(streamdetails.provider):
                 mass.create_task(music_prov.on_streamed(streamdetails, seconds_streamed))
 
-            if not streamdetails.loudness:
-                # send loudness analyze job to background worker
-                # note that we only do this if a track was at least been partially played
-                mass.create_task(analyze_loudness(mass, streamdetails))
-
 
 async def resolve_radio_stream(mass: MusicAssistant, url: str) -> tuple[str, bool]:
     """