The Theoretical Noise Floor by Bit Depth Calculator shows the dynamic range and quantisation noise floor for any audio bit depth, so you can see exactly what 16-bit, 24-bit, and 32-bit buy you. It is built for audio engineers, podcasters, and anyone choosing a recording format.
How it works
Digital audio stores each sample as one of 2^n levels, where n is the bit
depth. The gap between levels is the quantisation step, and the noise it
introduces sets the theoretical dynamic range. For a full-scale sine wave the
signal-to-quantisation-noise ratio is:
dynamic range (dB) = 6.02 × bits + 1.76
The simpler engineer’s rule drops the constant and uses 6.02 dB per bit, which gives the familiar 96 dB for 16-bit and 144 dB for 24-bit. The calculator shows both, plus the noise floor in dBFS, which is simply the negative of the dynamic range — the level at which quantisation noise sits below full scale.
Why each bit matters
Adding one bit halves the quantisation step, doubling resolution and gaining about 6.02 dB. So 16-bit reaches roughly 98 dB, 20-bit about 122 dB, and 24-bit about 146 dB of theoretical range.
Why 24-bit is enough
Human hearing spans roughly 120 dB from threshold to pain, and no microphone, preamp, or converter has a self-noise floor cleaner than about 130 dB. 24-bit’s ~146 dB already buries the analog noise floor, so extra bits cannot lower audible noise — the electronics dominate. 32-bit float is used for its vast headroom, which prevents clipping during capture and mixing, not for quieter recordings. Capture in 24-bit, mix in 32-bit float for safety, and deliver in whatever the target format requires.
Bit depth comparison table
| Bit depth | Dynamic range (approx) | Noise floor (dBFS) | Typical use |
|---|---|---|---|
| 8-bit | 49.9 dB | −49.9 dBFS | Vintage samplers, voice memos |
| 16-bit | 98.1 dB | −98.1 dBFS | CD audio, consumer playback |
| 20-bit | 122.2 dB | −122.2 dBFS | Some pro DAT formats |
| 24-bit | 146.2 dB | −146.2 dBFS | Studio recording and mastering |
| 32-bit int | 194.3 dB | −194.3 dBFS | Rarely used in audio; theoretical |
| 32-bit float | ~1,528 dB effective range | Effectively limitless | DAW mixing buses |
Values use the full formula 6.02 × bits + 1.76. The 32-bit float row is illustrative — its effective range comes from the floating-point exponent, not a simple quantisation model.
Why 8-bit sounds “lo-fi” and 16-bit sounds clean
The audible difference between 8-bit and 16-bit is easy to hear because 8-bit has only 256 possible amplitude levels. Quantisation distortion at that coarseness is clearly audible as a grainy, buzzy quality — sometimes called “bit crush.” CD-quality 16-bit has 65,536 levels, giving nearly 98 dB of range, which exceeds the dynamic range of most listening environments and masks the noise floor comfortably. That is why CD audio sounds clean even though modern studio recording uses 24-bit.
Dithering and bit reduction
When reducing from 24-bit to 16-bit for a CD master, a process called dithering adds a tiny amount of shaped noise before rounding. Counter-intuitively, this low-level noise masks the harsh quantisation distortion of the rounding step, making the perceptual quality better than an undithered truncation. The noise added by dithering is well below the audible threshold in a normal listening environment, but it means the theoretical noise floor of a dithered 16-bit file sits slightly above the raw −98 dBFS figure. This tool shows the theoretical floor without dither; in practice, a well-dithered 16-bit master sounds audibly better than a truncated one.