Dithering in Recording

Dithering adds controlled low-level noise to audio signals before reducing bit depth. This seemingly counterintuitive process replaces harsh quantization distortion with benign noise, improving the perceived quality of bit-reduced audio.

Why Dithering Exists

Reducing bit depth, such as converting from 24-bit to 16-bit for CD distribution, requires truncating or rounding amplitude values. Without dithering, this process introduces quantization distortion, particularly audible on quiet passages.

The distortion manifests as harsh, unmusical artifacts that correlate with the signal. Unlike random noise, this correlated distortion draws attention and sounds unpleasant.

Dithering adds random noise before truncation. This noise decorrelates the quantization error from the signal, converting obvious distortion into benign background noise.

When to Apply Dithering

Dithering applies when permanently reducing bit depth from higher to lower resolution. Converting a 24-bit master to 16-bit for CD release represents the most common application.

Each file should be dithered only once at the final export stage. Dithering multiple times accumulates noise unnecessarily.

Internal DAW processing at 32 or 64-bit float doesn’t require dithering since the full resolution is maintained. Dithering applies only at the final bit depth reduction.

Types of Dither

Flat dither adds white noise evenly across the frequency spectrum. This simplest approach effectively masks quantization distortion with minimal processing.

Noise-shaped dither concentrates the added noise at frequencies where human hearing is least sensitive, typically above 4 kHz. This psychoacoustically informed approach maintains equivalent distortion masking while reducing perceived noise.

Different noise-shaping curves optimize for various applications. Aggressive shaping minimizes perceived noise but may cause problems if the audio is further processed.

Application in Practice

Most DAWs include dithering options in their export dialogs. Selecting 16-bit export should typically engage dithering automatically or offer it as an option.

Dedicated dithering plugins provide more control over dither type and noise shaping. Products from companies like iZotope, Goodhertz, and others offer specialized dithering algorithms.

Placing the dither plugin last in the master chain ensures it processes the final signal before bit depth reduction. Position matters since processing after dithering can negate its benefits.

Hearing Dithering Effects

The difference between dithered and undithered audio is most apparent on quiet, isolated signals. A decaying reverb tail or solo instrument fadeout reveals quantization distortion that dithering eliminates.

In the context of full mixes, the difference may be subtle or inaudible. However, the artifacts from undithered truncation can become apparent on certain playback systems or during quiet passages.

A/B comparison of dithered versus undithered exports reveals the difference. Listening to the final moments of a fade or isolated quiet instruments shows the effect most clearly.

Common Mistakes

Dithering multiple times during successive exports accumulates noise. Each dithering pass adds its noise layer, eventually becoming audible.

Applying dithering when not reducing bit depth serves no purpose. A 24-bit to 24-bit export doesn’t need dithering since no truncation occurs.

Processing audio after dithering can modulate the dither noise, creating artifacts. Dither should be the absolute final process before export.

Bit Depth Workflow

Record at 24-bit to capture full dynamic range with headroom for gain staging. Maintain 24-bit throughout mixing and processing.

Export working bounces and stems at 24-bit without dithering to preserve full quality for further processing.

Apply dithering only at the final export stage when creating distribution files at 16-bit. This single application of dither at the end maintains optimal quality throughout the production chain.

Dithering for Different Destinations

CD mastering requires 16-bit output with appropriate dithering. Standard flat dither or moderate noise shaping works well for general music distribution.

Streaming services accept various bit depths but often process files internally. Providing 24-bit files when accepted lets the platform handle any bit depth reduction.

Files intended for further professional processing should remain at 24-bit without dithering. The next processing stage doesn’t need accumulated dither noise.