What Is Dithering in Audio? And Why Bit Depth Alone Isn’t Enough

Every time you bounce audio from a higher bit depth to a lower one, something small gets lost. I observed it many times.  

Not enough to hear on the first listen. But enough that mastering engineers have been using a specific fix for it since the earliest days of digital audio. 

That fix is dithering. 

Dithering is one of the least understood parts of the mastering chain because most producers never set it manually. Their tools handle it silently at the export stage, so the concept stays invisible. 

In this post, I’ll explain what dithering does, why bit depth alone can’t produce a clean digital master. Also, you’ll know which tools are actually doing when you bounce a final file.

Table of contents

• A Quick Refresher on Bit Depth
• What Actually Happens When You Convert Bit Depth
• What Dithering Actually Is
• The Types of Dither
• When Dithering Actually Matters
• What Your Mastering Tool Actually Does
• Common Dithering Mistakes
• FAQ

A Quick Refresher on Bit Depth 

Bit depth determines how many volume levels a digital audio sample can represent. 

• 16-bit captures 65,536 possible levels. This is the CD standard and the format most streaming platforms accept for lossless delivery. 
• 24-bit captures 16.7 million possible levels. This is the professional standard for recording, mixing, and mastering. 
• 32-bit float provides essentially unlimited dynamic range for internal DAW processing. 

Higher bit depth means more precision, not more frequency range. Sample rate handles frequency. Bit depth handles the fineness of each individual sample. Our guide on sample rate vs bit depth covers the full walkthrough if you want the deeper foundation. 

The point that matters for the rest of this post: 24-bit gives you far more precision per sample than 16-bit. When you convert between them, some precision has to go somewhere. 

What Actually Happens When You Convert Bit Depth

When you bounce a 24-bit mix to a 16-bit master (for CD delivery or lossless streaming), every sample has to round from 16.7 million possible levels down to 65,536. That rounding process is called quantization, and every rounded sample introduces a tiny error called quantization error. 

At high signal levels, this error is inaudible. The difference between one 16-bit level and the next is so small relative to the overall volume that your ears can’t detect it. 

At low signal levels, quantization error becomes audible. Specifically on fade-outs, reverb tails, and quiet passages, the rounded samples produce a distinct low-level distortion called quantization distortion. It sounds like a gritty texture on the tail of the sound. 

The problem isn’t the noise itself. Noise, we can filter out mentally. The problem is that quantization distortion is harmonically related to the audio signal, which means your brain doesn’t filter it out. It hears it as part of the music, and it recognizes something as wrong. 

This is the specific problem dithering solves. 

What Dithering Actually Is

Dithering is the deliberate addition of a very small amount of random noise to your audio before bit depth conversion. 

Adding noise to solve a problem sounds counterintuitive, but the math works. The added noise decouples the quantization error from the audio signal itself. Instead of the error creating harmonically-related distortion (which your ear identifies as wrong), the error becomes indistinguishable from the added noise (which your ear identifies as background). 

Your brain filters low-level noise automatically. It doesn’t filter harmonic distortion. So exchanging harmonic distortion for a tiny amount of noise is a genuine improvement, even though on paper you’re adding something to your audio rather than removing something. 

Dithering is applied at the final stage of bit depth conversion. If you’re going from 24-bit to 16-bit for a CD master or streaming delivery, dithering happens once, right at the end of your processing chain. iZotope’s technical guide to dithering demonstrates this visually with spectrograms if you want to see what the difference looks like on a meter. 

The Types of Dither

Most producers will never manually pick between these, but understanding them clarifies what your tools are doing. 

TPDF (Triangular Probability Density Function) dither (The industry standard)  

Uses noise with a triangular amplitude distribution, generated by summing two independent random noise sources. The critical property: TPDF dither eliminates quantization distortion completely without introducing noise modulation, meaning the noise floor stays perfectly constant regardless of what the audio signal is doing. This is the safe default for almost every situation. 

Noise-shaped dither  

TPDF dither with a filter applied that pushes the noise into frequency ranges above 15 kHz, where human hearing is less sensitive. The dither noise is still there, but you can’t hear it as well. Common implementations include Pro Tools’ POW-r family and iZotope’s MBIT+. Noise shaping increases the perceived dynamic range of a 16-bit master significantly. 

RPDF (Rectangular Probability Density Function) dither  

Simpler than TPDF, generated from a single random noise source. Leaves some signal-correlated noise modulation, meaning the noise floor subtly shifts with the music. Not recommended for final mastering. Included here because some older tools still offer it as an option. 

For modern music at 16-bit, TPDF with noise shaping is the professional standard. For 24-bit output, plain TPDF is sufficient because the noise floor is already so low that shaping produces no audible improvement. 

When Dithering Actually Matters

Practical rules for when to care about this. 

You should never dither more than once. Dithering is a permanent step in the sense that the noise gets locked into the signal. If you dither at the mastering stage and then dither again during distribution encoding, the second dither adds noise on top of noise. Once is correct. Twice is a mistake. 

Dithering must be the absolute last process in your chain. Any processing applied after dithering (EQ, compression, limiting, gain change, sample rate conversion) will undo the dither’s statistical properties and reintroduce quantization distortion. The correct order is: all processing, true peak limiting, then dithering, then export. 

What Your Mastering Tool Actually Does

Modern mastering platforms handle dithering automatically at the final bit depth conversion stage. 

When you upload a mix and get back a mastered file, dithering has been applied appropriately for the target bit depth. You don’t set it manually. You don’t choose the algorithm. The platform applies whatever is right for the output format. 

This isn’t laziness on your part. It’s the correct workflow.  

Manual dithering choices only matter in very specific scenarios (classical mastering to Blu-ray, archival work at ultra-low levels) that don’t apply to modern indie release workflows. 

The one thing you should verify: your DAW’s export settings and your mastering platform’s output specs. If you’re bouncing a 24-bit master from your DAW and uploading to a mastering service that outputs 16-bit files, dithering happens once at the mastering stage. That’s correct. If you dither at both ends, you’ve made a mistake that’s easy to fix by bouncing your DAW output at higher bit depth and letting the mastering process handle the final reduction. 

Our guide on how to master your track in 5 steps covers the broader mastering workflow that dithering fits into. 

Common Dithering Mistakes

Three mistakes that come up in real workflows. 

• Dithering twice. Bouncing a dithered 16-bit file from your DAW, then having a mastering service or plugin dither it again. Set your DAW bounce to the highest bit depth (24-bit or 32-bit float) and let the mastering process apply dithering only once at the final stage. 

• Adding a dithering plugin unnecessarily. If you’re not reducing bit depth, dithering adds noise for no benefit. Only apply dithering at the final bit depth conversion. 

• Processing after dithering. Applying any gain change, EQ move, or limiting after the dither stage destroys the dither’s statistical properties. Dithering must be the last step. If you notice a limiter or fader trim after your dithering plugin in your master chain, move the dither to the end. 

For a broader look at getting your file formats right, our guide on MP3 vs WAV vs FLAC covers the delivery-stage decisions that pair with dithering. 

Okay, Let’s wrap it  

Dithering is one of those technical concepts that matters conceptually but rarely requires manual attention in modern workflows. Understanding what it does, and why your tools handle it for you, is the whole point. 

You can’t hear the difference between a well-dithered master and a poorly-dithered one on most playback systems, but the professional standard exists for good reason. Fade-outs sound cleaner. Reverb tails feel smoother. Quiet passages don’t develop a gritty texture. These are small improvements, but they compound over the length of a track and across a full release. 

Bounce your mixes at high bit depth. Let mastering handle the final conversion. Trust that the tools built for this specific problem are applying the right dither at the right stage. 

Remasterify handles dithering automatically at the output stage. Upload a 24-bit or 32-bit float mix, and the mastering process delivers a properly dithered file at the bit depth your release needs. No manual configuration required. 

Dithering in Audio Related FAQs

1. What is dithering in audio? 

Dithering is the deliberate addition of a very small amount of random noise to audio before reducing its bit depth. The noise decouples quantization error from the audio signal, converting harmonic distortion (which your ear hears as wrong) into a barely audible background noise (which your brain filters out). Dithering is applied at the final stage of mastering, right before the file is exported at a lower bit depth. 

2. When do I need to use dithering? 

You need to dither whenever you reduce bit depth in your final output. The most common situation is bouncing a 24-bit mix down to a 16-bit master for CD or lossless streaming delivery. You should also dither when converting from 32-bit float down to 24-bit or 16-bit. You do not need to dither when bouncing intermediate mixes at the same bit depth or when exporting at 24-bit for further processing. 

3. What is the difference between TPDF dither and noise-shaped dither? 

TPDF (Triangular Probability Density Function) dither uses noise with a triangular amplitude distribution to completely eliminate quantization distortion. Noise-shaped dither takes TPDF and adds a filter that pushes the noise into frequency ranges above 15 kHz, where human hearing is less sensitive. For 16-bit output, noise-shaped TPDF is the professional standard. For 24-bit output, plain TPDF is sufficient because the noise floor is already so low. 

4. Should dithering be the last step in mastering? 

Yes. Dithering must be the absolute last process in your mastering chain. Any processing applied after dithering, including EQ, compression, limiting, gain change, or sample rate conversion, will undo the dither’s statistical properties and reintroduce quantization distortion. The correct chain order is: all processing, true peak limiting, then dithering, then export. 

5. Do I need to dither if my mastering tool handles it automatically? 

No, and you shouldn’t. Dithering should be applied only once during the final bit depth reduction. If your mastering platform handles dithering at the output stage, adding a dithering plugin in your DAW export creates a double-dither situation that adds noise on top of noise. Bounce your DAW output at the highest bit depth (24-bit or 32-bit float) and let the mastering process handle the final conversion.