sebastjan
  • Joined on 2026-04-20
sebastjan pushed to main at sebastjan/reels-app 2026-04-29 09:59:23 +02:00
af3c933c78 Robust language detection + anti-hallucination
sebastjan pushed to main at sebastjan/reels-app 2026-04-29 09:35:03 +02:00
c870d80726 Fix: extend clip if ends mid-vocal (no chorus cut-off), DejaVu Sans font (supports SLO/HR/BS chars), auto-upgrade to medium Whisper model for Slavic languages
sebastjan pushed to main at sebastjan/reels-app 2026-04-29 09:07:33 +02:00
5d5e169f9d Disable Whisper VAD filter — was dropping vocal segments in songs creating gaps in subtitles
sebastjan pushed to main at sebastjan/reels-app 2026-04-29 08:55:44 +02:00
a04811bdc9 Add Claude LLM analysis: sends full transcript to Claude API for true song structure understanding (refrain detection across all repetitions, not just local heuristic)
sebastjan pushed to main at sebastjan/reels-app 2026-04-29 08:33:56 +02:00
e072eec362 Fix: handle Whisper transcribe failure for instrumental-only audio (fallback to empty transcript)
sebastjan pushed to main at sebastjan/reels-app 2026-04-29 08:23:45 +02:00
33a138af9e Fix: force native Python bool/float for JSON serialization (numpy types)
sebastjan pushed to main at sebastjan/reels-app 2026-04-29 08:21:38 +02:00
8512076b91 Major: smart selection pipeline (analyze.py) + audio fade + multi-lang auto-detect
sebastjan pushed to main at sebastjan/reels-app 2026-04-29 08:09:30 +02:00
81edd24ca3 Subtitles: smaller font 56px (was 84), higher position MarginV=400, side margins 80px for safe zone
sebastjan pushed to main at sebastjan/reels-app 2026-04-29 07:59:39 +02:00
ba787744a6 Subtitles: cap chunk duration at 2.5s, split long lines into multiple time slices for faster reels pacing
sebastjan pushed to main at sebastjan/reels-app 2026-04-28 20:09:57 +02:00
e001387a89 Subtitles: convert SRT to ASS directly with PlayResY=1920 for predictable scaling instead of unreliable force_style
sebastjan pushed to main at sebastjan/reels-app 2026-04-28 19:40:51 +02:00
28d933c916 Subtitles: UPPERCASE + position lower (MarginV=320 for 1080x1920) + bigger font
sebastjan pushed to main at sebastjan/reels-app 2026-04-28 19:33:14 +02:00
d36893bf2d FIX CRITICAL: reload job dict after find_chorus update so reframe gets new start/duration values
sebastjan pushed to main at sebastjan/reels-app 2026-04-28 19:28:13 +02:00
15ef4888a1 Debug: log exact clip.py cmd in job + clip.py logs run_clip args
sebastjan pushed to main at sebastjan/reels-app 2026-04-28 19:17:13 +02:00
bc3fe1f9d4 Add explicit FFmpeg trim command logging + duration verification
sebastjan pushed to main at sebastjan/reels-app 2026-04-28 18:57:50 +02:00
8eaef029e2 Find chorus: weight repetitive short phrases (like 'Ohne dich x5') as strong chorus signal
sebastjan pushed to main at sebastjan/reels-app 2026-04-28 18:55:55 +02:00
c17578521a Fix find_chorus: RMS energy parser was broken (no pts_time available), now syntheses timestamps; energy weight x10 (refren je glasnejši)
sebastjan pushed to main at sebastjan/reels-app 2026-04-28 18:45:17 +02:00
64e8854cea Track mode: more sensitive face detection + longer smoothing window
sebastjan pushed to main at sebastjan/reels-app 2026-04-28 18:32:30 +02:00
400f6dbb6d Fix: limit FFmpeg crop expression to 20 sample points (was overflowing 4KB limit)
sebastjan pushed to main at sebastjan/reels-app 2026-04-28 18:29:43 +02:00
bf7ced5c7b Reset upload form also after failed jobs (so next upload works)
sebastjan pushed to main at sebastjan/reels-app 2026-04-28 18:22:43 +02:00
2e337ff079 Fix: shutil import was inside finally block, causing NameError when shutil.move was called