In this guide
Dither adds controlled low-level noise before reducing digital bit depth, changing correlated quantization distortion into noise. Decide on the final format first and check whether your export already applies it.
Audio dithering adds controlled low-level noise before reducing bit depth. It randomizes quantization error, trading some correlated low-level distortion for a small noise floor. A common example is preparing a 16-bit PCM WAV from a higher-resolution mix.
Dither is an export decision, not a noise-removal effect or a way to restore clipped audio. Its effect is most relevant at very low levels, such as quiet fades, and may be inaudible in ordinary listening. Keep the higher-resolution original and compare rather than assuming any conversion will sound better.
Why does reducing bit depth need attention?
Digital PCM samples represent amplitude using a finite set of values. When a signal is quantized to a lower bit depth, values must be rounded to the available steps. Without suitable dither, the resulting error can be related to the signal and produce distortion, especially at low levels. Dither randomizes that error before rounding, at the cost of added low-level noise.
The Audacity manual explains the randomization, and iZotope’s guide illustrates the difference on a quiet fade. This does not mean every audible hiss or distortion in a recording was caused by quantization.
When should you use dither?
| Situation | Practical choice | Check |
|---|---|---|
| Final 16-bit PCM WAV from a higher-resolution mix | Consider an appropriate 16-bit dither at the final bit-depth reduction. | Check whether your editor or export already applies it; avoid an unnecessary second dither stage. |
| Intermediate file for further editing | Keep a suitable higher-resolution working file where the next application supports it. | Follow the receiving engineer’s requirements instead of creating a 16-bit delivery file prematurely. |
| Floating-point export | Follow the editor’s documentation; fixed-point bit-depth reduction is a different operation. | Confirm the actual format and any processing applied during export. |
| MP3 converted to WAV | Choose the WAV format needed for the next step. | Neither WAV nor dither restores information lost in the MP3 encode. |
| Unmodified 16-bit file exported at the same resolution | Check whether the workflow actually requantizes the audio. | Audacity notes that a same-depth export without prior higher-precision operations has no benefit from dither. |
Dither is typically applied once at the final reduction to the target fixed-point depth. Applications can apply it automatically or expose different settings. Read the export documentation and avoid stacking a dither plug-in on top of an already dithered export by accident.
Bit depth is not sample rate
Bit depth describes the resolution used for each PCM sample; sample rate describes how many samples are stored per second. Changing 24-bit audio to 16-bit is not the same operation as changing 48 kHz to 44.1 kHz. A workflow may involve both, but dither addresses quantization at bit-depth reduction rather than choosing the sample rate.
For example, a 24-bit, 48 kHz master delivered as 16-bit, 44.1 kHz PCM needs both a sample-rate conversion and a bit-depth decision. Use an editor that documents the conversion and final dither, then inspect and listen to the actual deliverable. The WAV properties guide explains how to check the stored format.
A practical 16-bit WAV export checklist
- Finish the edit first. Check timing, gain, processing and the required delivery format. Keep the high-resolution project or master.
- Set the target format. Choose the requested PCM bit depth, sample rate and channel count. A WAV file is a container and does not by itself specify one bit depth.
- Check the dither setting. If export reduces to 16-bit PCM, select a documented dither option appropriate to that target, or confirm automatic behavior. Avoid applying it twice.
- Export and inspect. Open the file in another player or editor, check the beginning, quiet fade and ending, and verify its stored properties. Compare at matched playback level.
- Retain the original. If another revision is needed, return to the higher-resolution project rather than repeatedly processing a 16-bit delivery copy.
The local WAV file inspector can report stored sample rate, bit depth and channels for a supported WAV. It reads file properties; it cannot prove whether a file has been dithered or whether an export sounds transparent.
What are dither and noise shaping?
Dither is the low-level noise used to randomize quantization error. Noise shaping changes how some of that noise is distributed across frequencies to make it less perceptible in particular listening conditions. Total noise and peaks depend on the method; aggressive shaping is not automatically best. iZotope’s dither documentation describes the tradeoff.
Do not confuse dither with a noise gate or denoiser. A gate reduces level in quiet intervals, and denoising attempts to reduce unwanted recorded noise. Dither deliberately adds noise at quantization for a different reason. It cannot repair clipping, remove room echo or separate a voice from music.
What about Saturalabs audio tools?
Different tools have different export paths. The local audio equalizer, for example, exports WAV as 16-bit PCM and does not offer a dither control. If a delivery specification requires a particular dither method, keep the original and use an editor that provides the required bit-depth and dither options for the final export. The MP3 versus WAV guide covers working and delivery formats.
Common questions
Will dither make a recording louder or clearer?
It is not a loudness or clarity enhancer. It changes the character of quantization error during bit-depth reduction and adds low-level noise.
Should I dither every time I export?
Not automatically. Identify whether the export reduces to a lower fixed-point depth and whether the software already applies dither. Keep higher-resolution working files when appropriate.
Can a WAV inspector tell whether dither was applied?
Not from bit-depth metadata alone. The stored format does not record a reliable yes-or-no history of dither processing.
Can dither fix an MP3 or a clipped recording?
No. It does not recover detail lost to lossy encoding or restore peaks already clipped during capture.