keep audio skip consistent with playback

Clips were skipped even when the range would be served video-only, and seeking into a skipped span landed on later footage. The skip now only applies when every row in the range carries audio, and seeks inside a zero-duration span land on the next clip.

frigate/api/media.py

@@ -64,6 +64,7 @@ from frigate.util.recording_coverage import (
     plan_clip,
     recording_overlap,
     resolve_coverage,
+    sequence_needs_audio,
 )
 from frigate.util.time import get_normalized_tz_name
 
@@ -624,14 +625,14 @@ async def recording_clip(
 
 
 def _build_vod_clip(
-    row: Any, start: float, end: float, shared: bool
+    row: Any, start: float, end: float, needs_audio: bool
 ) -> tuple[dict[str, Any], int] | None:
     """Build one nginx-vod clip dict + duration (ms) for a recording row trimmed to [start, end).
 
     Realization comes entirely from the shared plan_clip, so the coverage
     endpoint's realized timelines match this manifest by construction.
     """
-    plan = plan_clip(row, start, end, shared)
+    plan = plan_clip(row, start, end, needs_audio)
 
     if plan.skipped:
         return None
@@ -690,6 +691,7 @@ async def _vod_response(
         null_audio_glitches(resolve_coverage(camera_name, start_ts, end_ts)),
         stream_preference,
     )
+    needs_audio = sequence_needs_audio(spans)
 
     durations: list[int] = []
     clips: list[dict[str, Any]] = []
@@ -707,7 +709,7 @@ async def _vod_response(
             row.end_time,
             row.duration,
         )
-        built = _build_vod_clip(row, span_start, span_end, len(spans) > 1)
+        built = _build_vod_clip(row, span_start, span_end, needs_audio)
 
         if built is None:
             continue

frigate/test/test_recording_coverage.py

@@ -477,6 +477,31 @@ class TestVodAudioCutShort(CoverageDbTestCase):
         clips = self._clips(1005.0, 1020.0)
         assert len(clips) == 2
 
+    def test_video_only_range_keeps_clip(self):
+        # a video-only sub row serves the whole range without audio, so
+        # nginx-vod never compares audio tracks
+        self._insert(
+            "m1",
+            1000.0,
+            1010.0,
+            "main",
+            has_audio=True,
+            audio_codec="aac",
+            audio_rate=16000,
+            keyframes=self.KEYFRAMES,
+            audio_duration=2.112,
+        )
+        self._insert(
+            "s1", 1010.0, 1020.0, "sub", has_audio=False, keyframes=self.KEYFRAMES
+        )
+        clips = self._clips(1005.0, 1020.0)
+        assert [c["path"] for c in clips] == ["/tmp/m1.mp4", "/tmp/s1.mp4"]
+        assert all(c["tracks"] == "v" for c in clips)
+        timeline = realized_timeline(
+            resolve_coverage("front_door", 1005.0, 1020.0), None
+        )
+        assert timeline[0]["duration"] > 0
+
     def test_sole_clip_is_kept(self):
         # alone in its sequence the clip plays video-only, so keep it
         self._insert_pair(2.112)

frigate/util/recording_coverage.py

@@ -341,7 +341,18 @@ class ClipPlan:
     first_key_frame_offset_ms: int = 0
 
 
-def plan_clip(row: Any, start: float, end: float, shared: bool = False) -> ClipPlan:
+def sequence_needs_audio(spans: list[list[Any]]) -> bool:
+    """Whether nginx-vod will require every clip in these spans to carry audio.
+
+    A range mixing audio-bearing and audio-less rows is served video-only,
+    and a single clip has nothing to mismatch against.
+    """
+    return len(spans) > 1 and all(row.has_audio is not False for row, *_ in spans)
+
+
+def plan_clip(
+    row: Any, start: float, end: float, needs_audio: bool = False
+) -> ClipPlan:
     """Plan one nginx-vod clip for a recording row trimmed to [start, end).
 
     The single source of truth for clip realization: the vod mapping
@@ -349,7 +360,7 @@ def plan_clip(row: Any, start: float, end: float, shared: bool = False) -> ClipP
     to the frontend, so the playhead model matches the playlist by
     construction rather than accumulating drift at each hand-off.
 
-    shared marks a clip that sits in a sequence with other clips.
+    needs_audio comes from sequence_needs_audio for the clip's range.
     """
     min_duration_ms = 100  # Minimum 100ms to ensure at least one video frame
     max_duration_ms = MAX_SEGMENT_DURATION * 1000
@@ -404,7 +415,7 @@ def plan_clip(row: Any, start: float, end: float, shared: bool = False) -> ClipP
     # nginx-vod then counts no audio track in this clip and fails the whole
     # sequence on the track count mismatch
     if (
-        shared
+        needs_audio
         and row.has_audio
         and row.audio_duration is not None
         and (clip_from or 0) >= int(row.audio_duration * 1000)
@@ -465,7 +476,7 @@ def realized_timeline(
     clip is skipped reports duration 0.
     """
     spans = build_spans(intervals, stream)
-    shared = len(spans) > 1
+    needs_audio = sequence_needs_audio(spans)
 
     return [
         {
@@ -474,7 +485,7 @@ def realized_timeline(
             "duration": plan.duration_ms,
         }
         for row, span_start, span_end, _ in spans
-        for plan in (plan_clip(row, span_start, span_end, shared),)
+        for plan in (plan_clip(row, span_start, span_end, needs_audio),)
     ]
 
 

web/src/utils/videoUtil.ts

@@ -61,6 +61,12 @@ export function calculateSeekPosition(
       return false;
     }
 
+    // skipped clips have no playlist media, so a seek inside one lands on
+    // the next clip that plays
+    if (segment.duration === 0) {
+      return true;
+    }
+
     // playlist duration exceeds wall length when the clip carries
     // keyframe back-snap lead-in
     const wallLength = segment.end_time - segment.start_time;