Initializing... drag & drop files here
Supports: MKV
Pull the soundtrack out of a Matroska video and save it as a standalone .opus file. Opus is the IETF's royalty-free audio codec, defined in RFC 6716 (September 2012), and it is what Discord, WhatsApp, YouTube, and WebRTC voice all run on. It stays clean at bitrates where MP3 falls apart, so a lecture capture or a stretch of dialogue lands at a fraction of the size. The video stream is discarded and the audio is re-encoded — pick a bitrate and you control the size-versus-fidelity trade directly.
.mkv onto the page or click "Add Files" to browse. Queue several videos to extract them all with the same settings in one pass..opus file. Files are uploaded over an encrypted connection, processed on our servers, and deleted automatically after a few hours — no sign-up, no watermark.These are Xiph.Org's own recommended settings for Opus, not a rule of thumb we invented. Opus is unusually flexible — RFC 6716 covers everything from 6 kbit/s narrowband speech to 510 kbit/s stereo music — so the right answer depends entirely on the content you extracted.
| Content you pulled out of the MKV | Channels | Xiph's recommended bitrate |
|---|---|---|
| VoIP / call recording | Mono | 10–24 kbps |
| Audiobook, podcast, lecture capture | Mono | 24 kbps (fullband) |
| Music for streaming | Stereo | 64–96 kbps |
| Music for storage / keeping | Stereo | 96–128 kbps |
| A 5.1 film mix kept as-is | 5.1 | 128–256 kbps |
| A 7.1 film mix kept as-is | 7.1 | 256–450 kbps |
Xiph notes that Opus at 128 kbps VBR is "pretty much transparent" for music, so there is rarely a reason to go higher when the source was itself lossy AAC or AC-3.
If you choose Variable Bitrate, the dropdown label is a range but the encoder aims at a single figure inside it. These are the targets our pipeline actually requests.
| Variable Bitrate option | Target the encoder aims for |
|---|---|
| 6k–24k | 15 kbps |
| 24k–40k | 32 kbps |
| 48k–64k | 56 kbps |
| 64k–96k | 80 kbps |
| 96k–128k | 112 kbps |
| 128k–160k | 144 kbps |
| 160k–192k | 176 kbps |
| 192k–256k | 224 kbps |
| 256k–320k | 288 kbps |
| 320k–510k | 415 kbps |
It re-encodes, always. Our server decodes whatever the MKV holds — usually AAC, AC-3, DTS, FLAC, or E-AC-3 — and runs it through the Opus encoder. There is no stream copy on this path, so even an MKV that already carries an Opus track goes through a fresh encode rather than being lifted out byte-for-byte. Match or slightly exceed the source bitrate if the MKV was already Opus, so the second generation costs you as little as possible.
Because Opus has no 44.1 kHz mode. The codec's native rates are 8, 12, 16, 24, and 48 kHz, and Xiph documents that input at those rates is internally converted to 48 kHz anyway. We checked our own encoder path: an unsupported rate is snapped to the nearest supported one before the encode runs, so choosing 44100 Hz produces a 48 kHz file. If you specifically need 44.1 kHz audio, Opus is the wrong container for the job — use MKV to MP3 or extract to WAV instead.
Not necessarily the first one. Matroska routinely holds a 5.1 mix, a stereo downmix, a commentary track, and several language dubs side by side. With no explicit stream mapping, ffmpeg's documented automatic selection picks "the stream with the most channels" for audio, breaking ties by lowest stream index — so the 5.1 mix usually wins even when the stereo track is listed first. If you need a specific dub or commentary, isolate that track in a Matroska editor before uploading.
It stays multichannel. Opus supports up to 8 channels through its channel-mapping families, and our encoder does not force a downmix when Audio Channel is Original — so a 5.1 film mix comes out as 5.1 Opus. That is worth knowing for two reasons: it needs a much higher bitrate than a stereo track to sound right (Xiph suggests 128–256 kbps for 5.1), and plenty of phones and browsers will fold it down on playback anyway. Set Audio Channel to Stereo if you want a predictable two-channel file.
.opus file play on the device I care about?Almost certainly on anything current, and almost certainly not on anything old. Per caniuse, Opus reaches about 96% of global browser usage; Chrome, Firefox, Edge, and Android have supported it for years, desktop Safari has had partial support since 11, and iOS Safari reached full support at 17.4 with further improvement in 18.4. VLC, mpv, foobar2000, and Discord all handle it natively. The weak spots are car head units, older smart TVs, iTunes, and hardware players — for those, MKV to MP3 is the safer target.
It is set by the bitrate you choose and the duration, not by the size of the MKV. The arithmetic is simply bitrate divided by 8: a 64 kbps extraction runs 8 kilobytes per second, so an hour of audio is roughly 28 MB, and the same hour at 32 kbps mono is roughly 14 MB. A 12 GB Blu-ray rip and a 700 MB web rip of the same film produce near-identical Opus files. Trim the source or drop the bitrate if you need something smaller.
They carry the same codec — the difference is the wrapper. .opus is a bare Ogg-encapsulated Opus stream, which is what general-purpose players, Discord, and command-line tools expect. MKV to WEBA puts the same Opus audio inside an audio-only WebM container, which is what HTML5 <audio> and WebM-based pipelines expect. If you are not building for the web, stay on .opus.
It is measurable and it is largest exactly where extracted dialogue lives. Opus was designed to cover speech and music in one codec, and its lead over MP3 is widest below about 96 kbps — an Opus file at 64 kbps is competitive with an MP3 at well over twice that rate. At 128 kbps and above the gap narrows because both are approaching transparency. Choose MP3 only when the destination cannot decode Opus.
Your MKV is uploaded over an encrypted connection, converted on our servers, and both the upload and the extracted audio are deleted automatically after a few hours. There is no sign-up, no watermark, and files are never shared or made public. If you only need a section of a long video, trimming the source first with the video cutter means a much shorter upload — the audio cutter takes audio files, not video.