What is audio bit depth?
For integer PCM, bit depth describes the number of bits used to represent each sample value. It affects quantization resolution and the theoretical noise floor/dynamic range of the digital representation.
Useful numbers
| Integer PCM | Approx. ideal quantization SNR for a full-scale sine | Typical context |
|---|---|---|
| 16-bit | ~98 dB | CD audio, distribution |
| 20-bit | ~122 dB | Historical/professional converters and files |
| 24-bit | ~146 dB | Recording, production, hi-res distribution |
The analogue electronics of real converters usually have a much higher noise floor than the mathematical limit of 24-bit PCM.
Storage bits and valid bits can differ
WAVEFORMATEXTENSIBLE can store, for example, 20 valid bits inside a 24-bit sample container. This is why “24-bit storage” and “24 bits of meaningful converter precision” are not always the same statement.
MP3/AAC/Opus do not preserve a PCM bit depth
A lossy codec may accept 16- or 24-bit PCM and decode back to a PCM representation, but “24-bit MP3” is not equivalent to 24-bit PCM or FLAC. The encoded signal is represented by the lossy codec's own transform/quantization model.
See 24-bit vs 16-bit.