Sample rate and bit depth explained: what 44.1 kHz and 16-bit actually mean

Every digital recording is described by two numbers — a sample rate like 44.1 kHz and a bit depth like 16-bit — and they are the source of a decades-long argument about what humans can hear. The physics is settled and worth knowing, because it tells you when a conversion is harmless and when it isn't; the sample rate converter does the conversion, and this explains what it's doing.

Sampling: slicing time

Sound is a continuous pressure wave. A digital recorder measures it at regular instants — samples — and stores each measurement as a number. Play the numbers back through a converter and, given two conditions, the original wave is reconstructed exactly. The conditions are the two numbers: how often you sample (rate) and how precisely you measure each sample (depth).

Sample rate and the Nyquist limit

The sampling theorem (Nyquist, Shannon) says a signal can be perfectly reconstructed if it contains no frequency above half the sample rate. Human hearing tops out around 20 kHz (less with age), so a rate a little above 40 kHz captures everything audible; CD's 44.1 kHz was chosen in 1979 to leave room for a filter above 20 kHz and to fit the video-tape recorders used for mastering. 48 kHz is video's standard for the same reason plus tidier maths against frame rates. Frequencies above half the rate must be filtered out before sampling or they fold back as false tones (aliasing) — which is why converting from a high rate to a low one is a filter followed by resampling, not just dropping samples.

Bit depth: the noise floor

Bit depth sets how finely each sample is measured. 16 bits give 65,536 levels; the rounding error between levels is quantisation noise, a hiss about 96 dB below full scale — quieter than any listening room. 24 bits push it to 144 dB, below the thermal noise of the electronics. Bit depth does not add "detail" in the sense people imagine; it lowers the noise floor and widens the dynamic range. That matters when recording (headroom for a quiet passage that's later boosted) and when processing (each edit adds rounding), which is why studios work at 24-bit and deliver at 16. Reducing depth should be done with dither (How dithering works — fewer colours that look like more) so the rounding becomes benign noise.

Is hi-res audio audible?

96 or 192 kHz at 24-bit is valuable for recording and mixing — headroom and gentler filters. For listening, controlled blind tests have repeatedly failed to show that people can distinguish a properly made 44.1/16 file from the hi-res master. The honest position: hi-res delivery does no harm and costs bytes; the audible differences between releases come from mastering, not sample rate. Lossy codecs are a different question — MP3 at low bitrate is easily audible (How MP3 compression works, and which bitrate you can actually hear).

Which settings to use

• Recording and editing: 48 kHz, 24-bit. Room to work.
• Music delivery: 44.1 kHz, 16-bit with dither — the CD standard, transparent.
• Video: 48 kHz; mixing 44.1 into a 48 project causes resampling anyway.
• Speech and podcasts: 44.1 or 48; the codec's bitrate matters far more than the rate.
• Converting: down is fine with a good filter; up adds nothing and just makes files larger.

Audio to WAV shows any file's rate and depth; the bitrate calculator turns them into file sizes.