RemuxForge: sync audio across MKV releases that don't line up

Remux mode

Same episode, two edits. One audio track that fits.

The Italian release of this episode has a little extra footage at six points, and is missing half a second at the end. Merged as-is, its audio slips further out of sync at every one of them. Deep Analysis finds them, and RemuxForge rebuilds the audio around them.

Episode 1 · 00:43:07

Source audio Italian audio Extra footage, cut Missing, filled from Source

#	Type	Source	Language	Duration
1	Cut	00:00:00.000	00:00:00.000	42 ms
2	Cut	00:07:04.341	00:07:04.383	1168 ms
3	Cut	00:13:21.759	00:13:22.969	1210 ms
4	Cut	00:21:22.447	00:21:24.867	1084 ms
5	Cut	00:28:23.367	00:28:26.871	876 ms
6	Cut	00:38:54.623	00:38:59.003	626 ms
7	Insert	00:43:07.001	00:43:12.007	542 ms

The real EditMap of this episode, found in 2 min 8 s. Positions are to scale; the width of each operation is enlarged so you can see it.

EditMap editor with synced frame previews, waveforms and spectrograms, to check or fix any operation by hand.
Inserts can be filled from the Source audio instead of silence, so a gap doesn't sound like a hole.
Audio conversion, 24→16 bit and peak normalization when tracks have to be re-encoded.
Whole seasons at once: episodes are matched by name, analysed in parallel, merged with MKVToolNix.

How synchronization works

Remux workspace

Direct merge

The audio already lines up.

instant

Frame-sync

Off by the same amount for the whole episode.

one analysis pass

Speed correction

Drifts steadily: right at the start, seconds off at the end. PAL/NTSC.

re-encodes the audio

Deep Analysis

Right in some places, wrong in others. Different edits.

slowest

Manual delay

You already know the exact offset.

free

Split mode

Cut on the frame you want, not the keyframe you get.

A segment that starts mid-GOP can't just be copied: its first frames are predicted from outside the segment. RemuxForge re-encodes only those, puts the original parameter sets back, and copies the rest byte for byte. The frame count is checked twice.

Split by chapters per episode, an explicit chapter pattern, split points, ranges, trims or manual segments.
Snap to the nearest keyframe for a pure stream copy, or cut exactly where you ask.
VFR timing, chapters, audio and subtitles carried into every segment. HEVC and AVC.

Split mode

Segment editor

Metadata mode

A whole library, one set of rules.

Write a preset once: an ordered pipeline of IF/THEN rules over container, video, audio and subtitle tracks. Run it on a folder and read the simulated result before a single byte changes.

SET container_title = {[file_stem]:NormalizeSpaces()}

SET audio_title = {[audio_language]:ToUpper()} [audio_format] [audio_channels_label]

IF audio_language = ita THEN audio_default exclusive

Field conditions, cross-track comparisons, track counts and OR groups.
Templated values, exclusive default flags, track removal, Matroska tags.
Bulk rename with stacked methods and a preview of every new name.

Writing presets

Preset editor

Screenshots

The rest of the interface.

Same UI in the desktop app and in the browser. Scroll sideways, click to enlarge.

Split workspace — Split · plan, coverage and segment map

Metadata workspace — Metadata · current data and simulated result

Advanced rename dialog — Bulk rename · preview before applying

Download

Get it. It's free.

Desktop app for Windows and macOS, WebUI for any machine or headless server, Docker image, and a CLI for scripts. Same engine in all of them.

Windows x64
macOS x64 · ARM64
Linux x64 · ARM64
Docker x64

Latest release

Needs MKVToolNix, ffmpeg (downloaded automatically on Windows and Linux; brew install ffmpeg-full on macOS) and MediaInfo CLI.

docker run -d \
  --name remuxforge \
  -p 5000:5000 \
  -e REMUXFORGE_DATA_DIR=/data \
  -v /path/to/config:/data:rw \
  -v /path/to/media:/media:rw \
  draknodd/remuxforge:latest

Paths typed in the WebUI are resolved inside the container.

# Windows
RemuxForge.Web.exe

# Linux / macOS
chmod +x RemuxForge.Web
./RemuxForge.Web

Then open http://localhost:5000. The port is configurable.

# Italian audio into an English release, constant offset
RemuxForge.Cli --mode remux -s "D:\Series.ENG" -l "D:\Series.ITA" -t ita -d "D:\Output" -fs

# A disc file into episodes, by chapter groups
RemuxForge.Cli --mode split --source "disc1.mkv" --pattern "5,5,5,6" --output-dir "D:\Output"

Every option is in the CLI reference.

Documentation

Written for the WebUI, with real cases.

Start

Remux

Split and Metadata

Worked examples

Reference

Under the hood

Open the wiki

Support

Free, and staying that way.

RemuxForge is open source under GPLv3. If it saves you an evening of nudging delays by hand, a coffee helps keep it going. Found a bug? Open an issue with the command line, tool versions and the relevant MediaInfo or ffmpeg output.

Buy me a coffee Open an issue