Support selected video clip audio

movmaker.py

@@ -49,6 +49,7 @@ class MovieData:
     place: str = ""
     description: str = ""
     captions: dict[str, str] | None = None
+    video_audio: set[str] | None = None
 
 
 @dataclass
@@ -587,6 +588,12 @@ def parse_data_file(path: Path | None) -> MovieData:
             data.place = value
         elif low in {"description", "desc", "synopsis"}:
             data.description = value
+        elif low.endswith(" audio"):
+            file_key = key[:-6].strip()
+            if file_key and value.lower() in {"1", "yes", "true", "on"}:
+                if data.video_audio is None:
+                    data.video_audio = set()
+                data.video_audio.add(file_key)
         elif key:
             data.captions[key] = value
         i += 1
@@ -649,6 +656,20 @@ def normalize_clip(item: MediaItem, out: Path, width: int, height: int, fps: int
     run(cmd)
 
 
+def media_has_audio(path: Path) -> bool:
+    try:
+        out = capture([
+            tool_path(FFPROBE, "ffprobe"), "-v", "error",
+            "-select_streams", "a:0",
+            "-show_entries", "stream=index",
+            "-of", "csv=p=0",
+            str(path),
+        ])
+        return bool(out.strip())
+    except Exception:
+        return False
+
+
 def audio_metadata(path: Path) -> tuple[str, str]:
     try:
         out = capture([
@@ -770,6 +791,7 @@ def build_video(
     clips: list[Path],
     durations: list[float],
     audio: Path | None,
+    clip_audio_paths: list[Path | None],
     output: Path,
     data: MovieData,
     creation_date: datetime,
@@ -801,6 +823,10 @@ def build_video(
         cmd += ["-i", str(clip)]
     if audio:
         cmd += ["-stream_loop", "-1", "-i", str(audio)]
+    clip_audio_input_start = len(clips) + (1 if audio else 0)
+    for clip_audio in clip_audio_paths:
+        if clip_audio is not None:
+            cmd += ["-i", str(clip_audio)]
 
     filters: list[str] = []
     n = len(clips)
@@ -936,13 +962,13 @@ def build_video(
             end = total_duration
         elif idx == 0:
             start = intro_duration
-            end = intro_duration + transition_offsets[0]
+            end = intro_duration + transition_offsets[0] + fade / 2
         elif idx == n - 1:
-            start = intro_duration + transition_offsets[idx - 1] + fade
+            start = intro_duration + transition_offsets[idx - 1] + fade / 2
             end = total_duration
         else:
-            start = intro_duration + transition_offsets[idx - 1] + fade
+            start = intro_duration + transition_offsets[idx - 1] + fade / 2
-            end = intro_duration + transition_offsets[idx]
+            end = intro_duration + transition_offsets[idx] + fade / 2
         if end <= start:
             continue
         next_label = f"stamp_left_{idx}"
@@ -956,6 +982,48 @@ def build_video(
         f"fontsize={stamp_size}:fontcolor=white:borderw=1:bordercolor=black[{final_label}]"
     )
 
+    audio_mix_labels: list[str] = []
+    if audio:
+        audio_index = len(clips)
+        audio_fade_out_start = max(0.0, total_duration - audio_fade)
+        filters.append(
+            f"[{audio_index}:a:0]atrim=0:{total_duration:.3f},asetpts=PTS-STARTPTS,"
+            f"afade=t=in:st=0:d={audio_fade_in:.3f},afade=t=out:st={audio_fade_out_start:.3f}:d={audio_fade:.3f}[bg_audio]"
+        )
+        audio_mix_labels.append("[bg_audio]")
+    clip_audio_input = clip_audio_input_start
+    for idx, clip_audio in enumerate(clip_audio_paths):
+        if clip_audio is None:
+            continue
+        if n == 1:
+            start = intro_duration
+            end = total_duration
+        elif idx == 0:
+            start = intro_duration
+            end = intro_duration + transition_offsets[0]
+        elif idx == n - 1:
+            start = intro_duration + transition_offsets[idx - 1] + fade
+            end = total_duration
+        else:
+            start = intro_duration + transition_offsets[idx - 1] + fade
+            end = intro_duration + transition_offsets[idx]
+        segment = max(0.0, end - start)
+        if segment <= 0:
+            clip_audio_input += 1
+            continue
+        fade_len = min(fade, segment / 2)
+        delay_ms = max(0, int(start * 1000))
+        label = f"clip_audio_{idx}"
+        filters.append(
+            f"[{clip_audio_input}:a:0]atrim=0:{segment:.3f},asetpts=PTS-STARTPTS,"
+            f"afade=t=in:st=0:d={fade_len:.3f},afade=t=out:st={max(0.0, segment - fade_len):.3f}:d={fade_len:.3f},"
+            f"adelay={delay_ms}:all=1[{label}]"
+        )
+        audio_mix_labels.append(f"[{label}]")
+        clip_audio_input += 1
+    if audio_mix_labels:
+        filters.append("".join(audio_mix_labels) + f"amix=inputs={len(audio_mix_labels)}:duration=longest:normalize=0[aout]")
+
     cmd += ["-filter_complex", ";".join(filters), "-map", f"[{final_label}]"]
     metadata = {
         "title": data.title,
@@ -973,11 +1041,8 @@ def build_video(
         if value:
             cmd += ["-metadata", f"{key}={value}"]
 
-    if audio:
+    if audio_mix_labels:
-        audio_index = len(clips)
+        cmd += ["-map", "[aout]", "-c:a", "aac", "-b:a", "192k", "-shortest"]
-        audio_fade_out_start = max(0.0, total_duration - audio_fade)
-        audio_filter = f"afade=t=in:st=0:d={audio_fade_in:.3f},afade=t=out:st={audio_fade_out_start:.3f}:d={audio_fade:.3f}"
-        cmd += ["-map", f"{audio_index}:a:0", "-af", audio_filter, "-c:a", "aac", "-b:a", "192k", "-shortest"]
     else:
         cmd += ["-an"]
 
@@ -993,7 +1058,7 @@ def main(argv: list[str] | None = None) -> int:
     parser.add_argument("--fade", type=float, default=3.0, help="crossfade duration in seconds")
     parser.add_argument("--audio-fade", type=float, default=10.0, help="music fade-out duration in seconds")
     parser.add_argument("--audio-fade-in", type=float, default=2.0, help="music fade-in duration in seconds")
-    parser.add_argument("--video-fade", type=float, default=1.5, help="video fade-in/fade-out duration in seconds")
+    parser.add_argument("--video-fade", type=float, default=1.0, help="video fade-in/fade-out duration in seconds")
     parser.add_argument("--resolution", type=parse_resolution, default=(1920, 1080), help="output resolution, e.g. 1920x1080")
     parser.add_argument("--fps", type=int, default=30, help="output frames per second")
     parser.add_argument("--preset", default="veryfast", help="x264 speed preset, e.g. ultrafast, superfast, veryfast")
@@ -1091,6 +1156,18 @@ def main(argv: list[str] | None = None) -> int:
         # Use probed normalized duration for accuracy.
         durations = [ffprobe_duration(out) for out in clips]
         captions = [data.captions.get(item.path.name, data.captions.get(str(item.path), "")) for item in media]
+        video_audio = data.video_audio or set()
+        clip_audio_paths = []
+        for item in media:
+            use_clip_audio = item.kind == "video" and (item.path.name in video_audio or str(item.path) in video_audio)
+            if use_clip_audio and media_has_audio(item.path):
+                clip_audio_paths.append(item.path)
+                console(f"Using video file audio: {item.path.name}")
+            elif use_clip_audio:
+                clip_audio_paths.append(None)
+                console(f"Video file has no audio stream: {item.path.name}")
+            else:
+                clip_audio_paths.append(None)
         clip_stamps = prepare_clip_stamps(media, data)
         audio = concat_or_mix_audio(audio_files, tmp / "music.m4a")
         if len(audio_files) > 1:
@@ -1099,7 +1176,7 @@ def main(argv: list[str] | None = None) -> int:
                 artist, title = audio_metadata(audio)
                 if artist or title:
                     console("Audio metadata: " + " - ".join(x for x in [artist, title] if x))
-        build_video(clips, durations, audio, output, data, first_media_date, captions, clip_stamps, fade, audio_fade, audio_fade_in, video_fade, width, height, fps, preset, crf, tmp)
+        build_video(clips, durations, audio, clip_audio_paths, output, data, first_media_date, captions, clip_stamps, fade, audio_fade, audio_fade_in, video_fade, width, height, fps, preset, crf, tmp)
 
     console(f"Done: {output}")
     return 0