8D audio. The sound circles your head.

There is no eighth dimension

It is worth saying plainly, because almost nothing written about this does. "8D audio" is automated panning. The track is moved slowly from one side of the stereo field to the other and back, and a small amount of reverb is added so the movement happens in a space rather than inside your skull. That is the whole technique. The number eight refers to nothing; it is a name that stuck.

That does not make the effect fake. It genuinely sounds like the music is circling you, the sensation is pleasant, and it is a real reason people listen to a song they already know. But it is not object-based spatial audio the way Dolby Atmos is, it is not binaural in the head-related-transfer-function sense, and it will not put a sound behind you. Anything claiming to add a dimension your speakers do not have is selling something.

Two consequences follow from that, and both matter more than any setting on this page.

It only works on headphones. On speakers, the two channels mix in the air before they reach you, so the panning collapses into a mild volume wobble. This is not a limitation of this implementation — it is what panning is. Anyone you send the file to needs headphones or they will wonder what you are talking about.

It works best on busy, wide tracks. A dense mix has things happening across the stereo field for the movement to carry. A sparse acoustic recording or a spoken-word file mostly sits in the middle, and moving the whole thing left and right reads as a fault rather than an effect.

The three controls, and what they actually change

The period is how long one full circuit takes. Ten seconds is the common choice and is slow enough that your ear reads it as travel. The interesting threshold is at the fast end: somewhere below about five seconds per circuit the brain stops tracking a moving source and starts hearing an amplitude effect, and the result sounds like tremolo rather than motion. That is why the fastest option here is six seconds rather than two.

The width is how far toward each ear the sound travels. Full width takes it all the way, which is more dramatic and also more tiring — at the extremes one ear is doing most of the work, and over a whole song that is noticeable. The standard setting stops short of the extremes for that reason.

The room is the reverb behind the movement. Without it the sound moves along a line between your ears, inside your head. With it the movement appears to happen in a space around you, which is most of what makes the effect convincing. It is a convolution reverb built from a synthesised impulse, with independent noise for the two channels so the tail spreads rather than collapsing to the centre, a short pre-delay, and damping on the wet path so it darkens as it decays the way a real room does.

The panning is driven by an oscillator wired into the pan parameter, rather than by a series of scheduled value changes. That detail is invisible in the result when it is right and obvious when it is wrong: stepping between a handful of positions produces audible zipper artefacts on a sweep this slow.

A mono file is promoted to dual mono before anything else happens. Panning a mono source does move it, but with silence in the other channel the effect is a volume wobble rather than a rotation, which is the single most common way a homemade 8D edit comes out sounding broken.

Practical notes

Check it on headphones before you judge it. This sounds obvious and is the most common complaint about every tool that does this. On laptop speakers the effect is close to inaudible by design.

Level is preserved, not normalised. Adding a room raises the peak, so the output is scaled back only if it would otherwise clip. A quiet recording comes back at its original level rather than being pushed up, which means a voice memo does not return with its noise floor raised.

Start from the best source you have. The reverb and the panning both expose encoder artefacts that sat unnoticed in a static mix, because your ear tracks a moving source more closely than a stationary one. A 128 kbps MP3 will sound worse after this than it did before. If you have the lossless file, use it and choose WAV output.

Stems give a much better result than a stereo mix. The convincing version of this effect moves the instruments independently rather than swinging the whole mix as one object. That is beyond a single-file tool, but the stem splitter will separate vocal from instrumental locally, and running one part through here and leaving the other still is a noticeably better edit than moving everything together.

Tags do not carry across. The output is a new file. The tag editor writes title and artist back without re-encoding.

On uploading it. An 8D edit of a commercial track is still that track, and content matching identifies it — panning does not defeat fingerprinting. Expect a public upload to be claimed.