Fixing volume propagation settings through events.json, and fixed bug where failure to match the text on the end slide of a segment now issues a warning, and uses the last word + 2 seconds, rather than the full duration of the clip

gnommo/cli.py

@@ -3241,8 +3241,19 @@ def cmd_trim(
                         end_text = slide_texts.get(end_slide, "")
                         end_ts = _find_slide_end_in_transcript(words, end_text, verbose) if end_text else None
                         if end_ts is None:
-                            end_abs = total_dur
-                            end_note = f"end auto: S{end_slide} not found → to end {total_dur:.2f}s"
+                            # Slide-end text didn't match the transcript (e.g. the
+                            # narrator drifted from the manuscript). Fall back to the
+                            # last spoken word + tail — NOT total_dur, which includes
+                            # trailing silence and leaves the clip playing long.
+                            end_abs = min(words[-1].end + _TRIM_TAIL_OUT, total_dur)
+                            end_note = f"end auto: S{end_slide} not found → last word {words[-1].end:.2f}s +{_TRIM_TAIL_OUT:g}s → {end_abs:.2f}s"
+                            print(
+                                f"\n  ⚠️  {seg_id}: could NOT match the end of S{end_slide} in the transcript.\n"
+                                f"     The manuscript text for S{end_slide} likely drifted from what was\n"
+                                f"     spoken. Falling back to the last spoken word ({words[-1].end:.2f}s)\n"
+                                f"     + {_TRIM_TAIL_OUT:g}s instead of the raw file length ({total_dur:.2f}s).\n"
+                                f"     Check S{end_slide}'s text or set an explicit `end` for {seg_id}."
+                            )
                         else:
                             end_abs = min(end_ts + _TRIM_TAIL_OUT, total_dur)
                             end_note = f"end auto: S{end_slide} last word {end_ts:.2f}s +{_TRIM_TAIL_OUT:g}s → {end_abs:.2f}s"

gnommo/models.py

@@ -434,6 +434,9 @@ class VideoEvent:
     cutout: "CutoutDefinition"
     cutout_name: str = ""  # resolved cutout name (e.g. "fullscreen"), for display
     layer: str = "above"  # "above" = on top of slides; "below" = behind slides
+    # Effective audio volume for THIS occurrence: an events.json override if present,
+    # else the videos.json default. The renderer reads this (not video_source.volume).
+    volume: float = 1.0
     # Chunking v2 seam (see docs/chunking_v2.md): when a clip began before this
     # chunk's window, the render must seek into it so it resumes mid-clip instead of
     # restarting at the boundary. None = play from video_source.skip (the v1/default).

gnommo/parser.py

@@ -58,13 +58,15 @@ def _resolve_case_insensitive(path: Path) -> Path:
 # presentation fields materialized onto events.json and resolved per-event
 # (transformer.resolve_video_presentation). Global params (skip/zoom/volume/…) are not
 # yet overridable inline; unknown keys are ignored.
-_MARKER_OVERRIDE_KEYS = frozenset({"cutout", "layer", "end_on", "take"})
+_MARKER_OVERRIDE_KEYS = frozenset({"cutout", "layer", "end_on", "take", "volume"})
+# Override keys that are numeric (coerced to float).
+_MARKER_NUMERIC_KEYS = frozenset({"take", "volume"})
 
 
 def _coerce_marker_value(key: str, raw: str):
-    """Coerce an inline override value: `take` → float; everything else → string."""
+    """Coerce an inline override value: numeric keys → float; everything else → string."""
     v = raw.strip().strip('"').strip("'")
-    if key == "take":
+    if key in _MARKER_NUMERIC_KEYS:
         try:
             return float(v)
         except ValueError:

gnommo/renderer.py

@@ -1568,7 +1568,8 @@ def build_filter_complex(
             delay_ms = int(event.start_time * 1000)
             label = f"tvaud{i}"
 
-            vol = event.video_source.volume
+            # event.volume = events.json override if set, else the videos.json default.
+            vol = event.volume
             vol_filter = f",volume={vol:.2f}" if vol != 1.0 else ""
             filters.append(
                 f"[{video_idx}:a]atrim=0:{duration:.3f},"

gnommo/scaffold.py

@@ -30,9 +30,13 @@ from typing import Optional
 from .models import CAMERA_PRESETS
 from .transformer import MarkerTiming, resolve_video_presentation
 
-# Per-occurrence presentation fields materialized onto video events (atomic events.json,
-# GUI-ready). Round-tripped as overrides so a stored value drives render over the default.
+# Per-occurrence presentation fields ALWAYS materialized onto video events (atomic
+# events.json, GUI-ready). Round-tripped as overrides so a stored value drives render.
 _PRESENTATION_KEYS = ("cutout", "layer", "end_on", "take")
+# `volume` is materialized SPARSELY — only when actually overridden (inline/GUI/manual),
+# so the videos.json value keeps flowing as the default and only a real override pins it.
+# Round-tripping still carries it whenever present in the event dict.
+_EVENT_OVERRIDE_KEYS = _PRESENTATION_KEYS + ("volume",)
 
 EVENTS_FILE = "events.json"
 SCAFFOLD_FILE = "scaffold.json"
@@ -184,6 +188,10 @@ def derive_events(
                 e["end_on"] = pres["end_on"]
                 if pres["take"] is not None:
                     e["take"] = pres["take"]
+                # volume is sparse: written only when actually overridden, so the
+                # videos.json default keeps flowing until someone pins it here.
+                if t.overrides and "volume" in t.overrides:
+                    e["volume"] = pres["volume"]
         pd = _pause_duration(t.marker_id, videos)
         if pd:
             e["pause_duration"] = pd
@@ -327,7 +335,7 @@ def events_to_marker_timings(events: list[dict]) -> list[MarkerTiming]:
         eff = (n + e.get("adjustment", 0.0)) if n is not None else -1.0
         # Carry any stored presentation as overrides so render honors the atomic event
         # (e.g. a GUI edit) over the shorthand/videos.json default. Absent on old events.
-        overrides = {k: e[k] for k in _PRESENTATION_KEYS if k in e} or None
+        overrides = {k: e[k] for k in _EVENT_OVERRIDE_KEYS if k in e} or None
         timings.append(
             MarkerTiming(
                 marker_id=e["id"],

gnommo/transformer.py

@@ -97,6 +97,8 @@ def resolve_video_presentation(
     pause_narration = overrides.get(
         "pause_narration", video_source.pause_narration or 0.0
     )
+    # Volume defaults to the videos.json value; an events.json/inline override wins.
+    volume = overrides["volume"] if "volume" in overrides else video_source.volume
 
     return {
         "handle": handle,
@@ -105,6 +107,7 @@ def resolve_video_presentation(
         "end_on": end_on,
         "take": take,
         "pause_narration": float(pause_narration or 0.0),
+        "volume": float(volume if volume is not None else 1.0),
     }
 
 
@@ -1368,6 +1371,7 @@ def _extract_video_events(
         end_on = pres["end_on"]
         take = pres["take"]
         pause_narration = pres["pause_narration"]
+        volume = pres["volume"]
 
         if end_on == "take" and take is not None:
             end_time = start_time + take
@@ -1440,6 +1444,7 @@ def _extract_video_events(
                 cutout=cutout,
                 cutout_name=cutout_name,
                 layer=layer,
+                volume=volume,
                 skip_override=skip_override,
             )
         )