8D Audio Converter

Turn a song into 8D audio in your browser with a rotating left-right pan, adjustable speed and depth, plus optional reverb. Free, private, no upload.

🌐 Español

Drop your file here (.mp3, .wav, .m4a, .ogg, .flac)

🔒 Private by design: your files are processed locally in your browser and never uploaded to any server.

Written and maintained by Last updated

What the two dropdowns actually set in the filter chain

The name promises a dimension count that does not exist. What “8D audio” describes is an auto-pan: the whole mix is swept back and forth between the left and right channels on a slow, continuous sine cycle. Because headphones deliver each channel to one ear with no acoustic mixing in between, that sweep is heard as the music orbiting your head. It is a stereo illusion, not binaural or HRTF spatial audio, and no per-ear filtering is involved.

Underneath, this page builds a short FFmpeg filter chain and the two selects set one number each in it. Rotation speed sets the cycle frequency: 0.05 Hz for slow, 0.1 Hz for medium, 0.2 Hz for fast, which is one full rotation every twenty, ten or five seconds respectively. Effect depth sets the modulation amount on a zero-to-one scale: 0.6, 0.85 or 1. Depth is how far toward each side the signal swings, and at the maximum the sound spends real time nearly entirely in one ear.

Medium and medium is the pairing both selects open on, and the safest place to start. Fast plus strong is the setting that turns a novelty into motion sickness inside a minute.

A mono source is upmixed before the pan can do anything

There is a step before the pan that matters more than it sounds. Panning a mono signal achieves nothing audible, because both channels are already identical and moving level between them just moves the same sound around. So the chain begins by forcing a stereo channel layout, which upmixes a mono input to two channels first and is a harmless no-op on a track that was already stereo.

This is worth knowing if your source is a voice memo, a phone recording, or an old mono transfer. The effect will apply and the file will come back with a rotation on it, but it will be a rotation of one signal rather than of a stereo image, so it sounds emptier than the same treatment on a proper stereo mix. Wide, reverb-rich, well-separated productions are the ones where the illusion is strongest.

Rotating a track and checking it back

  1. Drop one audio file onto the box above, or use Choose a file. MP3, WAV, M4A, OGG and FLAC are accepted, one file per run.
  2. Set Rotation speed and Effect depth. Leaving both on Medium is the safe starting point if you have not done this before.
  3. Tick Add reverb (echo-based ambience) if you want a hazier wash behind the movement, or leave it off for a cleaner rotation.
  4. Press 8D Audio Converter and let the progress bar run, keeping the tab open.
  5. Download the MP3, which arrives with -8d appended to your original file’s name, and listen on headphones.

That last step is not optional advice. Checked on speakers, the left and right signals recombine in the room before they reach you and the whole illusion collapses into a plain left-to-right volume wobble.

The reverb tick box is an echo filter, and says so

The optional reverb is not a convolution reverb. It is FFmpeg’s echo filter set to a single soft, widely spaced repeat, which is the usual stand-in for an ambient wash in this genre of edit and the same approach the slowed-and-reverb preset on the audio pitch changer uses. It gets you close to the hazy sound that pairs with slowed edits. It will not match a real reverb plugin in a proper editor, and it is not trying to.

If the goal is atmosphere rather than rotation specifically, it is worth trying the neighbouring effects instead of stacking everything here: Change Audio Speed alters the tempo while leaving the pitch where it was, and Bass Booster lifts a low shelf if the bottom end needs weight before the pan starts moving it around. And if you want genuine two-tone binaural beats rather than a panned song, Binaural Beats Generator synthesises them from scratch instead of processing a file you upload.

Where the illusion falls flat

Three source types disappoint reliably. Anything already heavily side-processed, with instruments hard-panned to the extremes, fights the sweep instead of riding it. Dense mono-ish loudness-war masters have little stereo image left to move. And podcast or spoken-word audio becomes actively hard to follow, because a voice that drifts out of one ear reads as a fault rather than an effect.

Everything on this page runs through a WebAssembly build of FFmpeg inside your browser tab, on your own hardware, so an unreleased demo or a rough mix never gets uploaded anywhere to be processed. The rest of the effect pages sit together under audio tools.

See it in action

Screenshot of the 8D Audio Converter tool with sysfenix-sample.wav (517 KB) loaded, Rotation speed set to Medium, one rotation every ~10 seconds, Effect depth set to Medium, a clear rotation
8D Audio Converter mid-process: sysfenix-sample.wav (517 KB) loaded, Rotation speed set to Medium, one rotation every ~10 seconds, Effect depth set to Medium, a clear rotation.
Screenshot of the 8D Audio Converter result screen showing sysfenix-sample-8d.mp3 ready to download (72 KB, 86% smaller)
The finished result: sysfenix-sample-8d.mp3 ready to download (72 KB, 86% smaller). The download link is a local blob URL — the file never leaves your device.

Frequently asked questions

My 8D version lost the album artwork. Where did it go?

The command includes a flag that tells FFmpeg to ignore video streams, and an embedded cover image inside an MP3 or M4A is stored as exactly that, a single-frame video stream. So the picture is not carried across to the 8D file. If the artwork matters, re-attach it afterwards with any tag editor; the audio itself is untouched by that.

Which side does the rotation start on, and can I line it up with the beat?

You cannot choose either. The pan filter's two channel offsets are left at their defaults, which put the left and right channels half a cycle apart, and the cycle begins with the file rather than at some musical landmark. The upside is that it is deterministic, so the same song at the same settings always produces exactly the same motion. There is no phase, offset or beat-sync control on this page.

Can I apply the rotation to just the chorus instead of the whole track?

No. The filter chain is applied across the file from the first sample to the last, and there are no start and end fields here. The practical route is to cut the section you want out first and run that shorter file through this page, which produces the same result for that stretch and takes proportionally less time.

Why is the output always an MP3 even when I upload a WAV or FLAC?

The encoder is fixed to LAME at 192 kbps. The pan rewrites every sample, so there is no lossless copy path available whatever you upload, and a single output format keeps the result easy to share. It does mean a lossless source comes back lossy, so keep your WAV or FLAC master if you plan to do anything else with it later.

Can I run this on the audio of a video file?

Not directly. The dropzone takes .mp3, .wav, .m4a, .ogg and .flac, and any other extension is listed back as an unsupported file type without being processed. Extract the audio from the video first, run that through here, and put the 8D track back over the picture afterwards if that is where it is going.

How long does a full song take to process?

There is no honest fixed figure, because it depends on the track length and on how fast a single core of your device is. The FFmpeg build used here is the single-threaded one, since the multithreaded build needs response headers this site cannot set, so the work does not spread across your cores. The progress bar tracks FFmpeg's own position through the file. One extra thing to expect is that the first run of a session loads the FFmpeg core before any encoding starts.

Related tools