If you are asking what is dithering, here is the short answer: dithering is a tiny amount of carefully shaped noise added to audio when you reduce its bit depth (for example, from 24-bit down to 16-bit). It masks the harsh distortion that would otherwise appear in the quietest parts of the signal. Used at the right moment, it makes a lower-bit-depth file sound cleaner.
Quick answer
- Dither is low-level noise added when reducing bit depth.
- It hides quantisation distortion in quiet passages and fades.
- Apply it only at the final bit-depth reduction (e.g. exporting a 16-bit file).
- Apply it once, as the last step, never repeatedly.
Why dithering exists
When you lower bit depth, you reduce the number of values available to describe each sample. The audio has to be rounded to the nearest available value, a process called quantisation. In loud passages this rounding is harmless, but in very quiet passages and fade-outs it produces a correlated, granular distortion that the ear hears as harshness rather than gentle noise. Dithering replaces that ugly distortion with a smooth, far less objectionable noise floor. This sits alongside the concepts in our guide to sample rate and bit depth.
What is dithering doing?
By adding random noise just before the bit-depth reduction, dither decorrelates the quantisation error from the signal. Instead of distortion that tracks the music (which the ear notices), you get a constant, very low-level hiss that the ear largely ignores. Many dither algorithms also use noise shaping, which pushes that noise into frequency ranges where human hearing is less sensitive, making it even harder to notice.
A simple way to picture it
Imagine you can only describe a slow fade using whole numbers: 3, 2, 1, 0. Without dither, the signal jumps in hard steps from one value to the next, and those steps are correlated with the music, so your ear locks onto them as a gritty, buzzy edge on the tail of the note. Dither adds a faint random wobble before the rounding happens, so the signal spends part of its time on the higher value and part on the lower one. Averaged over time, that trade keeps the impression of the in-between levels alive even though the file itself only stores whole numbers. The cost is a tiny, steady hiss, which is far kinder to the ear than stepped distortion. This is why dither is sometimes described as trading a small amount of broadband noise for a large amount of perceived detail in quiet material.
When should you use dither?
The rule is simple: dither only when you reduce bit depth, and only once, as the very last step. Common cases:
- Exporting a 16-bit file (such as for CD) from a 24-bit project — dither.
- Bouncing a 24-bit master to 16-bit — dither on that export.
- Staying at 24-bit or 32-bit float — no dither needed.
If you are delivering 24-bit files (common for streaming and most mastering), you usually do not dither at all. Loudness and delivery formats are covered in how loud your master should be.
How to apply dither correctly
Place the dither as the final process in your master chain or in your DAW’s export dialog, after the limiter. If you are unsure what that last stage is doing, our guide to what a limiter does in mixing explains why nothing should sit between it and the dither. Set the dither’s target bit depth to match your output (16-bit for CD-quality). Crucially, apply it only once — re-dithering an already-dithered file adds noise on noise. Most DAWs and mastering tools include a dither option in the export settings or as a dedicated plugin (for example, the dither modules found in mastering suites and many of the best free mastering plugins).
Common dithering mistakes to avoid
Dither is one of those processes that quietly goes wrong when you are not paying attention. A few traps come up again and again:
- Dithering more than once. If you dither a 16-bit file, then later import it, edit it, and dither again on a second export, you stack noise on top of noise. Each pass should be the last and only one for that file.
- Dithering when you are not reducing bit depth. Adding dither while exporting a 24-bit file from a 24-bit project does nothing useful and only adds a touch of noise. No bit-depth reduction means no need for dither.
- Putting dither before the limiter. Anything you do after dithering — gain changes, limiting, further processing — alters the carefully placed noise and can re-introduce the very distortion you were trying to avoid. Dither belongs at the very end.
- Forgetting it entirely on a 16-bit export. The omission is hard to hear on a busy mix, but on sparse material, quiet intros, and long fades the stepped distortion becomes audible. If you are bouncing to 16-bit, dither is not optional.
- Obsessing over which algorithm to choose. The differences between modern noise-shaping options are subtle and far less important than simply applying dither once, in the right place. Pick a sensible default and move on.
Does dithering matter for everyday work?
For most home-recording and streaming-first projects, you will spend the majority of your time at 24-bit or 32-bit float, where dither never enters the picture. It becomes relevant mainly at the delivery stage, and only when a target format demands 16-bit. That makes dither a small, occasional decision rather than a constant concern — but when the moment arrives, getting it right is the difference between a clean fade-out and a faintly gritty one. Treat it as a deliberate final checkbox on any 16-bit export — right after you bounce your final file — and you will rarely think about it otherwise.
For the full mastering workflow this fits into, see what mastering is and the wider mixing and mastering hub.
Frequently asked questions
Do I always need to dither?
No. You only dither when reducing bit depth, such as bouncing a 24-bit project to a 16-bit file. If your output stays at 24-bit or 32-bit float, dithering is unnecessary.
Can I hear dither?
Rarely, and only if you turn it up very loud in a silent passage. Properly applied dither sits far below the music as a faint, smooth noise floor that is far less noticeable than the distortion it replaces.
Should I dither before or after the limiter?
After. Dither is the final step in the chain, applied at the point of bit-depth reduction, so it goes after the limiter and any other processing. Apply it only once.
Does noise shaping make dither better?
It can. Noise shaping moves the dither noise into frequencies where your ears are less sensitive, so the same amount of noise becomes harder to perceive. It is a refinement of plain dither rather than a different process, and for 16-bit delivery a noise-shaped option is usually a sensible default.
