Levels & Metering

What is Dithering?

Dithering is the process of adding a tiny amount of low-level random noise to an audio signal before reducing bit depth, which replaces harsh quantization distortion with a benign, inaudible noise floor.

How It Works

When you reduce the bit depth of an audio file — for example, from 24-bit to 16-bit for CD or from 32-bit float to 24-bit for a high-res download — you are reducing the number of possible amplitude values each sample can take. The quietest signals, which used the fine resolution of those extra bits, must now be rounded to coarser step values. This rounding (quantization) creates a specific type of distortion called quantization noise, which is correlated with the signal — meaning it changes in character depending on what is playing, producing an audible graininess or grittiness on quiet passages and fade-outs. Dithering solves this by adding a carefully designed random noise signal — typically at an extremely low level (around -90 to -96 dBFS for 16-bit) — before the bit depth reduction. This noise randomizes the rounding errors, converting the correlated quantization distortion into uncorrelated, constant-level noise that sounds like a very quiet, smooth hiss. Perceptually, this is a massive improvement: the ear easily accepts a tiny constant noise floor but is sensitive to the modulating artifacts of undithered quantization. Noise shaping takes dithering a step further by using psychoacoustic principles to push the dither noise into frequency ranges where human hearing is least sensitive — primarily above 10 kHz. This effectively makes the added noise even less audible without increasing its overall energy. Common noise shaping algorithms include POW-r (with three types of increasing aggression), MBIT+, and Apogee UV22HR. The choice matters most for 16-bit output; for 24-bit conversions, the noise floor is already so low that simple flat dither is sufficient.

Why It Matters for Your Mix

Dithering is one of those technical details that separates truly professional output from "close enough." The difference is most audible on quiet passages, fade-outs, and reverb tails — exactly the moments where listeners are paying close attention. Without dithering, a 16-bit fade-out of a piano recording will exhibit a grainy, metallic distortion as the signal drops into the lowest bits. With proper dithering, the same fade-out is smooth, clean, and natural, disappearing into a noise floor that is 96 dB below full scale — well below the ambient noise of any listening environment. The practical rule is straightforward: apply dither exactly once, at the very last stage of processing, whenever you reduce bit depth. If you are mastering to 16-bit for CD, dither goes on last. If you are exporting a 24-bit master from a 32-bit float session, dither at the export. If you are not reducing bit depth — for example, bouncing a 24-bit session to a 24-bit file — dithering is not needed because no quantization is occurring.

Common Mistakes

Applying dither multiple times

Dither should be applied once and only at the final bit depth conversion. Dithering multiple times (for example, on every bounce in a multi-stage workflow) accumulates noise with each pass. Set your dither plugin at the very end of the master bus chain and apply it only when rendering the final file.

Forgetting to dither when exporting to 16-bit

Many producers export 16-bit WAV files for CD or distribution without dithering, assuming the DAW handles it automatically. Most DAWs do not apply dither by default — you must add it yourself. The resulting quantization distortion is subtle but audible on quiet material, and it is entirely preventable.

Using noise shaping for 24-bit output

Noise shaping is designed for 16-bit output where the quantization noise floor is close to audible. At 24-bit, the noise floor is already 144 dB below full scale — far below human hearing. Applying aggressive noise shaping to 24-bit output is unnecessary and can introduce issues if the file is later converted to 16-bit.

How We Analyze This in Your Mix

RoastYourMix examines the noise floor characteristics of your uploaded audio to identify signs of proper dithering or its absence. We analyze the low-level signal behavior during quiet passages and fade-outs, looking for the spectral signature of quantization distortion (correlated noise) versus properly dithered output (uncorrelated broadband or shaped noise). Tracks submitted at 16-bit without dithering are flagged with specific recommendations.

See Dithering in Action

Upload your mix and see how dithering affects your track.

Get Your Mix Roasted

Frequently Asked Questions

Apply dither whenever you reduce bit depth, and only at that final conversion step. Common scenarios: mastering a 24-bit or 32-bit float mix down to 16-bit for CD, or exporting a 32-bit float session to 24-bit for distribution. If you are bouncing at the same bit depth as your session, dithering is not needed.

For 16-bit output, noise-shaped dither (POW-r Type 1 or 2) is generally the best choice because it pushes the noise into less audible frequency ranges. For 24-bit output, simple flat triangular (TPDF) dither is sufficient since the noise floor is already inaudible. When in doubt, TPDF dither is a safe, transparent choice for any bit depth conversion.

On a full, loud passage, you will not hear a difference. On quiet passages, reverb tails, and fade-outs at 16-bit, the difference between dithered and undithered audio is clearly audible — undithered audio has a grainy, rough quality, while dithered audio fades smoothly into a clean noise floor. Try it yourself: render a fade-out at 16-bit with and without dither and compare on headphones.

Related Terms

Clipping Normalization

Ready to Hear the Truth?

Upload your mix and get instant feedback. Free health score, frequency analysis, and actionable fixes.