12 May 2026 | 6 min read
Audio bit depth explained: 16, 24 and 32 bit float in plain language
Bit depth is not resolution in the photographic sense. It is headroom and noise floor, and confusing the two leads to a lot of bad advice.
Sample rate decides how often the signal is measured. Bit depth decides how precisely each measurement is stored. More bits does not mean more detail in the way more pixels means more detail. It means a lower noise floor and more room between your quietest usable sound and digital silence.
The numbers
- 16 bit: about 96 dB of dynamic range. The CD standard, fine for finished delivery.
- 24 bit: about 144 dB. The production standard, because it forgives conservative recording levels.
- 32 bit float: a moving exponent instead of a fixed range, effectively removing clipping at the converter stage on recorders that support it.
Why 24 bit is the working standard
On set you set levels for the loudest thing that might happen, not the average. That means dialogue often sits 20 dB or more below full scale. At 24 bit, the noise floor is so far down that this costs nothing. At 16 bit, the same conservative level starts to expose quantisation noise once you push gain in post.
What 32 bit float really does
On a recorder with dual converters, 32 bit float means a take that looks clipped can be pulled back down and still be clean, because the peak was never truncated. It does not fix a microphone overloading, a preamp distorting or a wind blast. It fixes the converter stage only.
Converting down safely
Going from 24 bit to 16 bit throws away the bottom of the range, so apply dither if the material is quiet or has long fades. Going from 32 bit float to 24 bit fixed requires setting gain first, because a float file can legally hold samples above zero dBFS that a fixed format cannot.
The export controls here let you choose bit depth explicitly, and the inspector shows you what the source actually is before you decide.