Bit Depth
How many bits each audio sample uses — and what that means for dynamic range and noise.
Bit depth is the number of bits used to encode each sample of audio. The standards you'll meet are 16-bit, 24-bit, and 32-bit (floating point). Each carries a different amount of dynamic range and noise.
It is not the same as sample rate, which is how many samples per second are taken. Bit depth is about each individual sample's resolution; sample rate is about how often samples are captured.
Quick Reference
| Bit depth | Dynamic range | Noise floor | Use case | |-----------|---------------|-------------|----------| | 16-bit | 96 dB | −96 dBFS | Streaming, CD, final delivery | | 24-bit | 144 dB | −144 dBFS | Recording, mixing, mastering source | | 32-bit float | ~1500 dB | Effectively zero | Working sessions, plugin processing |
What Bit Depth Means
Each audio sample is a number. The bit depth says how big that number can be.
- 16-bit: 65,536 possible values per sample.
- 24-bit: 16,777,216 possible values per sample.
- 32-bit: ~4 billion possible values, plus extended range via floating-point notation.
More possible values means more precision in capturing the input signal — and a lower noise floor where quantisation noise (the tiny error from representing analogue with discrete numbers) lives.
Dynamic Range
Each bit of depth gives you 6 dB of dynamic range:
- 16-bit × 6 dB = 96 dB
- 24-bit × 6 dB = 144 dB
The dynamic range is the gap between the loudest signal a system can encode and the noise floor below which signal becomes inseparable from quantisation noise.
For perspective:
- A whispered conversation: ~30 dB SPL
- A quiet bedroom: ~30 dB SPL
- Normal conversation: ~60 dB SPL
- A loud rock concert: ~110 dB SPL
- A jet engine at 30 m: ~140 dB SPL
The dynamic range of human hearing from threshold to pain is about 120 dB. So 24-bit captures the full range of human hearing with margin to spare. 16-bit captures everything except the very quietest noises.
When Bit Depth Actually Matters
Recording. Always 24-bit. The headroom matters because you don't always know how loud the signal will get, and a 24-bit recording with peaks at −18 dBFS still has a 126 dB usable dynamic range.
Plugin processing. Most modern DAWs use 32-bit float internally. This means plugin chains can boost and cut without introducing quantisation noise — even if a single plugin clips internally, the chain doesn't lose data.
Final delivery to streaming/CD. 16-bit, dithered. Dither adds a tiny amount of random noise to mask the quantisation distortion that occurs when you reduce bit depth.
Dither
When you reduce from 24-bit to 16-bit, you lose precision. Dither adds a small random noise that "smears" the quantisation error so it sounds like soft hiss instead of harsh distortion.
Most DAWs let you choose between several dither algorithms (POW-r, MBIT+, TPDF). The differences are subtle. Use whatever your DAW recommends; the important thing is to always apply dither when reducing bit depth.
32-bit Float
Two things make 32-bit float special:
-
Floating-point math. Numbers can be very small or very large without losing precision relative to themselves. A signal at −60 dBFS still has 24 bits of usable dynamic range available.
-
No clipping in the file. A 32-bit float file can hold signals far above 0 dBFS without distorting. If you accidentally render a clipped mix, you can pull the gain back and the audio is intact.
Modern field recorders (Zoom F3, Sound Devices MixPre-II 32-float) use 32-bit float capture so you can never clip a recording — useful for unpredictable sources like dialogue and explosions.
Common Mistakes
Recording at 16-bit. Old DAW templates sometimes default here. 24-bit gives you 8 more bits of headroom for free.
Skipping dither when reducing bit depth. Going 24-bit → 16-bit without dither produces audible quantisation distortion at low signal levels. Almost no professional bounce should skip dither.
Believing 24-bit is "audibly better" than 16-bit. For final playback, 16-bit is indistinguishable from 24-bit when properly dithered. The benefit of 24-bit is during recording and mixing — extra headroom and lower noise floor — not at the final playback.
Confusing bit depth and sample rate. They're orthogonal. A 96 kHz / 16-bit recording is technically possible, just not a common combination. 48 kHz / 24-bit is the modern recording standard.