How to Convert MKV to Text: Multi-Track Audio & OBS Recordings

The Short Version

Upload your MKV to a transcription tool and it works, with one catch: if your file has multiple audio tracks (common in OBS recordings, Blu-ray rips, and multi-mic setups), you need to know which track holds the audio you actually want. Run a quick ffprobe first, map the right track, and the rest is routine.

What MKV Is and Where Your Files Came From

MKV (Matroska Video) is the most flexible video container in common use. Unlike MP4, it places no hard limits on the number of audio tracks, subtitle tracks, or chapters inside a single file. That flexibility is why MKV is common in:

• OBS Studio recordings (recommended for multi-track work even though OBS 32 changed the default for new installs)
• Ripped DVDs and Blu-rays with multiple audio languages
• Anime and TV archives with original-language plus dubbed audio
• Multi-mic podcast setups where each microphone lands on its own track
• Conference recordings with separate speaker mic feeds
• Linux screen recorders like SimpleScreenRecorder and Kazam

For transcription, MKV's multi-track support is both an asset and a complication. You need to pick the right track before transcribing.

The OBS Default Changed in Version 32

The existing advice "OBS records MKV by default" is now outdated. OBS 30.2 introduced Hybrid MP4 (a crash-resistant fragmented MP4), and OBS 32.0 made Hybrid MP4/MOV the default container for new installs. If you installed OBS recently, you may already be getting Hybrid MP4 files, not MKV.

That said, MKV remains the recommended format for multi-track recording. Hybrid MP4's crash protection covers the single-track use case. Multi-track MKV is still the better choice when you want isolated audio per source, because:

• MKV supports up to 6 audio tracks in OBS (Hybrid MP4 typically records as a single combined stream)
• You can always remux MKV to MP4 via File > Remux Recordings without re-encoding
• Editors and transcription tools can then select the specific track they need

My take: if you are recording a podcast, interview, or conference with separate microphones, use MKV in OBS explicitly. If you are recording a single-source screengrab or game capture, Hybrid MP4 is fine and simpler.

What the Tracks in an OBS MKV Actually Contain

OBS documentation describes Track 1 as the default "all sources" combined mix. Tracks 2 through 6 are where you route isolated sources in Advanced Audio Properties. A typical multi-track OBS setup might look like:

• Track 1: Combined stereo mix of everything (default playback track)
• Track 2: Your microphone only
• Track 3: Desktop/game audio only
• Track 4: Communications (Discord, calls)

For transcription, Track 2 (mic only) is almost always what you want. Track 1 picks up keystrokes, game audio, and background noise alongside your voice.

Step 1: Inspect Your Tracks Before Uploading

Ten seconds with ffprobe saves you from transcribing the wrong language or a noise-heavy combined mix. Run:

ffprobe -v error -show_streams -select_streams a -of json input.mkv

This outputs each audio stream with its index, codec name, channel count, and language tag. For a typical Blu-ray rip you might see:

• Stream 0:1 - AC3, 6 channels, lang: eng
• Stream 0:2 - AAC, 2 channels, lang: spa
• Stream 0:3 - AC3, 2 channels, lang: fra (director commentary)

For a typical OBS multi-track MKV:

• Stream 0:1 - AAC, 2 channels (combined mix)
• Stream 0:2 - AAC, 1 channel (microphone)
• Stream 0:3 - AAC, 2 channels (desktop audio)

Identify the stream index before moving forward. Track indices are zero-based in ffmpeg (the second audio track is 0:a:1).

Step 2: Extract the Right Track

If your MKV has a single audio track, upload as-is. If it has multiple tracks, extract the one you need first. This also reduces file size dramatically. A 90-minute movie MKV might be 8 to 15 GB; the audio track alone is typically 100 to 300 MB.

Extract a specific audio track without re-encoding:

ffmpeg -i video.mkv -map 0:a:1 -c:a copy track2.m4a

Substitute 0:a:1 with 0:a:2 for the third audio track, and so on.

If the file is too large to upload because of video, strip video entirely:

ffmpeg -i in.mkv -vn -c:a copy audio.mka

The .mka extension is Matroska Audio, a container most transcription tools accept. You can also use .aac, .mp3, or .m4a depending on the source codec.

Step 3: Handle Codec Compatibility

MKV can carry codecs that trip up web-based transcription tools. The common ones:

AC3 and AAC work everywhere. No action needed.

Opus works everywhere. OBS recordings commonly use Opus.

FLAC works in most tools. It is lossless and large, but transcription services handle it.

DTS and DTS-HD are common in Blu-ray rips and often rejected by online tools. Convert first:

ffmpeg -i bluray.mkv -map 0:a:0 -c:a libmp3lame -b:a 192k audio.mp3

TrueHD (Dolby Atmos) is also a Blu-ray staple and has the same issue. Same fix, same command, just specify the right stream index.

Multi-channel (5.1, 7.1) audio can confuse speaker detection. For pure speech-to-text, downmix to stereo or mono:

ffmpeg -i input.mkv -map 0:a:0 -ac 2 -c:a aac output.m4a

Step 4: Upload and Set Language

Match the transcription language to the audio track you extracted. If you are transcribing the Spanish track, set Spanish in the tool. Language mismatch is the most common cause of garbled transcription output.

For video-to-text workflows, you can use the video to text tool on ConvertAudioToText. Drag the file in, set the language, and let it run. A 2-hour recording finishes in roughly 10 to 20 minutes depending on queue load.

If you want per-speaker accuracy on a multi-mic recording rather than relying on automatic diarization, see speaker diarization explained for when per-track extraction beats AI speaker splitting.

The Subtitle Track Shortcut (When It Applies)

MKV can carry embedded subtitle tracks alongside its audio. Two distinct types exist:

Text-based subtitles (SRT, ASS, SSA): These are stored as readable text streams. If your MKV already has accurate subtitles, you can extract them directly without running any transcription:

ffmpeg -i video.mkv -map 0:s:0 subtitles.srt

This is instant and free. It only works if the subtitles already exist and are accurate.

PGS subtitles (Blu-ray image subs): These are stored as bitmap images, not text. ffmpeg -map 0:s:0 will export a .sup file, not readable text. Converting PGS to SRT requires OCR software (Subtitle Edit's OCR module is the standard tool), and the output typically needs manual cleanup. For Blu-ray rips with PGS subs, re-transcribing the audio is usually faster and more accurate than running OCR on image bitmaps.

To check what subtitle type your MKV has:

ffprobe -v error -show_streams -select_streams s -of json input.mkv

Look at the codec_name field. subrip or ass means text-based. hdmv_pgs_subtitle means image-based bitmap PGS.

Per-Source MKV Workflows

OBS Recordings

For OBS multi-track MKV files, always extract the mic-only track (typically Track 2) before transcribing. The combined mix on Track 1 includes keyboard noise, game audio, and notification sounds that degrade accuracy.

If you recorded streams and want the voice over only, enable multi-track audio in OBS before your next recording session: Settings > Output > Recording > Audio Track checkboxes. Once set up, the MKV will separate every source.

Ripped DVDs and Blu-rays

DVD rips typically have multiple language tracks plus commentary. Language tags are your guide, but they are not always accurate. Spot-check with VLC by switching audio tracks manually before committing to transcription. For a movie research project, consider whether you want the original-language track (authoritative dialogue) or a dubbed track (may differ from the original script in phrasing and timing).

Director commentary tracks are often mistakenly transcribed when users forget to specify a track. Check with ffprobe first.

Conference and Multi-Mic Recordings

Multi-room and multi-mic recording systems can produce MKV with one track per microphone. For transcript quality, transcribing individual mic tracks separately beats relying on diarization applied to a noisy combined mix.

Extract each track, transcribe each separately, then merge the outputs in timestamp order. This produces clean speaker attribution without any speaker detection ambiguity. For more on when that tradeoff is worth the extra steps, see speaker diarization explained.

Linux Screen Recorders

OBS, SimpleScreenRecorder, and Kazam on Linux all produce MKV. Audio quality depends on the audio backend. PipeWire (the modern default on most current distributions) captures cleanly at 48 kHz. Older PulseAudio setups can introduce dropouts every few seconds during long recordings. If your MKV has periodic audio glitches, check whether a PipeWire migration on your system would fix it before blaming the transcription tool.

MKV vs Other Multi-Track Containers

MKV is the right container for content with multiple audio or subtitle versions. It is not the right choice if you need maximum compatibility with consumer devices or streaming platforms, which expect MP4. See supported audio formats for transcription for how different containers affect upload compatibility across tools.

Common MKV Transcription Problems

Wrong language in the transcript

Your tool defaulted to Track 1. If your MKV has English and Spanish audio, and you needed Spanish, re-extract with -map 0:a:1 and re-upload.

"Unsupported codec" error

Almost certainly DTS or TrueHD from a Blu-ray rip. Convert to MP3 or AAC with the ffmpeg command in the codec section above.

Language tag is wrong

Some encoders mistag tracks. A file tagged "eng" that plays Japanese audio is common in older rips. Do not trust the tag blindly. Verify by listening to a 30-second sample in VLC.

MKV chapter markers causing segment errors

Some tools trip on Matroska chapter markers embedded in the file. Strip them:

ffmpeg -i in.mkv -map 0 -map_chapters -1 -c copy out.mkv

File too large to upload

The video stream is why. A 2-hour movie at 1080p is 8 to 15 GB. The audio stream is 100 to 300 MB. Strip video first with the -vn command above.

If you want a fast no-signup path for a single MKV with no track complications, ConvertAudioToText accepts MKV uploads and processes them without an account for up to 30 minutes. The free signed-in tier gives 10 minutes per month; the Pro tier at $9.99/mo removes the cap entirely.

FAQ

Can I transcribe multiple audio tracks at once?

Most tools transcribe one track per job. For multi-track MKV files, run separate transcription jobs per track, then merge results in timestamp order. This is the cleanest approach for multi-speaker podcasts where each speaker has their own track and is more reliable than automatic speaker diarization.

Is MKV audio quality better than MP4?

The container does not affect audio quality. MKV can hold higher-quality codecs (FLAC, DTS-HD, TrueHD) than standard MP4, but a typical OBS MKV uses the same AAC that MP4 uses. Codec choice matters; container choice does not.

Can I transcribe MKV for free?

Yes. ConvertAudioToText accepts MKV files and processes up to 30 minutes without an account, or 10 minutes per month on a free plan. For longer files or multi-track per-speaker workflows, the Pro tier at $9.99/mo removes the limit entirely.

What about MKV files with no audio language tag?

Open in VLC and listen to a few seconds of each track under Audio > Audio Track. Or trust the order: in most movie rips, track 1 is the original language and subsequent tracks are dubs. Not guaranteed, but a reasonable starting point before running ffprobe.

Sources

• OBS Studio 32.0 Release Notes - confirmed Hybrid MP4/MOV became the default container in OBS 32
• Hybrid MP4 and Hybrid MOV Formats | OBS Knowledge Base - format details and crash-recovery mechanics
• Multiple Audio Track Recording Guide | OBS Knowledge Base - track assignment and Track 1 behavior
• Advanced Recording Guide and Multi Track Audio | OBS Knowledge Base - verified Track 1 as all-sources combined mix
• Standard Recording Output Guide | OBS Knowledge Base - MKV recommendation for multi-track recording
• Understanding PGS Subtitles | suptosrt.com - PGS as bitmap-based subtitle format
• Extracting Subtitles and Captions from Video Files with FFmpeg | Mux - text subtitle extraction workflow
• ConvertAudioToText Pricing - verified free tier (10 min/mo account, 30 min no-signup) and Pro plan ($9.99/mo annual)