15 articles about subtitles.
Going from a transcription API to SRT is where caption pipelines break. Fix timestamps, line wraps, and escaping, then burn them on with FFmpeg or one call.
The gpt-4o-transcribe retirement hits October 15, 2026 with no replacement listed. Two migration paths, plus the API chain that burns captions into video.
FFmpeg's codec not currently supported in container error is a subtitle mux failure. Fix it with -c:s mov_text, burn in bitmap subtitles, or switch to MKV.
The extract subtitles FFmpeg one-liner is easy. Finding the right track with ffprobe, handling PGS bitmaps, and fixing mov_text in MP4 is where jobs break.
Convert SRT to VTT without losing cue positioning: what FFmpeg's ASS intermediate silently drops, how to go back to SRT, and how to shift cues after a trim.
The TikTok safe zone is roughly 108 px top and 320 px bottom on 1080x1920. Convert them into FFmpeg subtitles MarginV and drawtext y= values that scale.
The FFmpeg whisper filter transcribes video natively in FFmpeg 8.0, but only in a build compiled with --enable-whisper. Command, models, and production gaps.
Auto translate subtitles video pipelines break at the burn-in, not the translation. Transcribe once, translate the SRT, render one hardsub per language.
The FFmpeg fontconfig error is a warning, not a failure: the job exits 0 and ships the wrong font. Fix it in Alpine, Debian slim, and distroless images.
Whisper FFmpeg burn in captions API: transcribe to SRT and burn subtitles into video with two HTTP calls. No GPU, no local FFmpeg build, no 25 MB cap.