Convert MKV to MOV

OBS Studio records to Matroska by default, and it is a considered choice rather than an accident. Matroska writes its structure continuously, so a capture that ends in a crash, a power cut or a frozen game is still playable up to the second it stopped. The same session recorded straight to MP4 would frequently be unrecoverable, which is the failure the default exists to prevent.

Stream archives, conference captures and anything produced by a recorder that expects to be interrupted arrive the same way, for the same reason. The cost is paid later, by whoever has to edit it: the container that survives everything is the container almost no editing application will import.

OBS can route game audio, a microphone, a browser source and a voice chat to different tracks in one recording, so that the edit can duck one under another rather than being handed a single mixed-down file. That structure is the most valuable thing in the MKV and it is the thing most converters destroy.

Here it survives. Every audio track the browser can decode is written into the MOV in the order the MKV held them, because the QuickTime container places no practical limit on how many it can carry. On a Final Cut or Premiere timeline they land as separate channels. This is the sharp difference between converting the video and extracting the audio: an MP3, a WAV or a FLAC out of the same MKV would keep exactly one of those tracks and silently discard the rest.

The video track is written as H.264 at High profile, with AAC audio. The level is not fixed: it is looked up from the frame size, so a 1080p recording comes out at level 4.0 and a 4K one at level 5.1. That is the pairing the browser’s own hardware encoder produces, and it is what every editor on a Mac imports without a transcode step. It is not ProRes, DNxHD or any other intermediate codec, because a browser has no encoder for those.

The practical difference is felt on the timeline rather than in the picture. A long-GOP delivery codec makes the machine work harder on every scrub and every frame-accurate trim, where an intermediate codec is designed to be cheap to decode at any point. For interview takes, screen segments and B-roll it is entirely workable. For a multi-camera edit of a two-hour stream, transcode the MOV to ProRes in Final Cut once and edit from that.

Before any track is converted, the browser is asked whether it can decode it. If the answer is no, the track is dropped and the picture continues without it, so the conversion succeeds and produces a MOV that plays with no sound at all. Nothing announces it at the moment it happens.

This never bites an OBS recording, which uses AAC. It bites the other kind of Matroska file — one taken off a disc, where the soundtrack is AC-3, E-AC-3, DTS or Dolby TrueHD. Those are cinema codecs with their own licensing and browsers ship no decoders for them. If the MKV you were sent came from a disc rather than a capture card, check the audio in the MOV before you build anything on it.

This conversion reads video tracks and audio tracks and nothing else. A subtitle track in the MKV is not declined, it is never seen — which means a file with burned-in captions and a file with switchable ones look identical to the converter and come out very differently. Chapter marks disappear the same way.

On an editing job that is usually the right outcome, because captions belong in the edit rather than in the source. Where it matters, MKVToolNix extracts every subtitle track out of an MKV in seconds without touching a frame, and an SRT is something Premiere and Resolve both import as a caption track directly.

Raise the quality band. The default is balanced, which is tuned for watching, and this is footage that will be re-encoded at least once more when the edit is exported. Every generation of compression is paid for by the one after it, and the high band is the cheapest insurance available on the page.

Leave the resolution alone unless there is a reason not to. Asking for more than the source has is ignored by design, so the control only ever reduces. Reducing on the way into an edit is a decision you cannot take back later, and unlike the quality band it saves nothing you will miss — the encoder does most of its work per pixel regardless.

VLC and media servers play Matroska without complaint, which is how these files circulate freely and why the refusal is always a surprise. Editing software is a different proposition: it needs to seek to an arbitrary frame, read the file backwards, and rely on a stable index, and the vendors have simply never shipped Matroska support for that.

Final Cut Pro will not import an MKV at all. Premiere Pro’s support has come and gone across versions and cannot be relied on for a delivery deadline. Resolve is the most tolerant of the three and still refuses particular codec combinations inside the container. Converting once, up front, is shorter than discovering the limit at the point where a render fails.

It runs in the browser tab, on your own processor, using the codecs the machine already has. Nothing carrying the recording is sent anywhere — which for footage that has not been released yet, or a client capture under an agreement, is the difference between a workable tool and one that cannot be used at all.

The free tier stops at 100 MB. A full stream archive is far past that, and the honest recommendation for a session-length file is a desktop encoder. What fits here is the practical middle: pulled takes, interview segments, a demo capture, the ten minutes of a stream that are actually going into the cut.