Digital Audio: Samples, Bits and Nyquist

Read first: SoundStudio theory section 2, Digital audio, and the table in section 14, File formats. The tutorial explains samples, sample rate, bit depth and Nyquist with a good diagram. Here we turn those ideas into decisions: what rate to record at, what to export, and what to stop worrying about.

A file is a list of numbers

A computer cannot store a smooth curve. So it measures the height of the wave over and over, very fast, and keeps the list. Each measurement is a sample. The sample rate is how many measurements per second: 48,000 of them is 48 kHz. The bit depth is how precisely each one is written down.

Think of a film camera. The sample rate is the frame rate. The bit depth is how many shades of gray each frame can show. A low frame rate misses fast motion. Too few shades and smooth skies turn into bands. Audio fails in the same places.

Nyquist: the ceiling

The rule that runs digital audio is short. A file can hold frequencies up to half its sample rate, and no higher. That half is the Nyquist limit. You need at least two samples per cycle, one for the top of the wave and one for the bottom, to pin down a frequency.

Both 44.1 and 48 kHz put the ceiling above 20 kHz, past the top of human hearing. For listening, they are equally good. The choice between them is about what system the audio is going into.

What happens if a frequency above the limit sneaks in? It does not disappear. It folds back and shows up as a false, lower tone that was never in the original. That is aliasing, and it sounds like a thin metallic whine that does not follow the music. Every converter has a steep filter in front of it to stop this, so you will rarely meet aliasing from a microphone. You can meet it from cheap plugins, extreme pitch shifting and badly done sample rate conversion.

Why 48 kHz for video and 44.1 for music

The CD rate is an accident of history. In the late 1970s, the only tape machines with enough capacity for digital audio were video recorders. Sony's PCM adaptors stored audio as patterns in a video signal, and 44.1 kHz was a rate that fit both PAL and NTSC line timing. The compact disc kept it. Music has lived at 44.1 kHz ever since.

Video and broadcast went with 48 kHz, which divides cleanly into common frame rates and became the professional standard. DVD and Blu-ray video use 48 kHz and its multiples. YouTube's upload guidance suggests 48 kHz audio as of October 2026. Most cameras and phones record video sound at 48 kHz by default.

Mixing rates is the real danger. Drop a 44.1 kHz music bed into a 48 kHz video project, and a good editor converts it for you. A careless chain might play it at the wrong speed instead: 48 divided by 44.1 is about 1.088, so the music would run nearly 9% fast and about a semitone and a half sharp. If a song suddenly sounds like the band had too much coffee, check the sample rate first.

Higher rates like 96 kHz have a place. They give time-stretch and pitch-shift tools more data to work with, and some engineers like the gentler filters. For a podcast, a radio spot or a town council recording, 96 kHz just doubles your file sizes.

Bit depth: the floor

Each bit of depth adds about 6 dB of range between the loudest possible sample and the noise floor of the format. The tutorial gives the numbers: 16-bit is about 96 dB, 24-bit about 144 dB. For reference, a quiet room plus a decent microphone preamp rarely beats 100 dB of real-world range, so 24-bit is more than any room can fill.

The useful rule is about recording, not playback. Record at 24-bit. It lets you leave plenty of headroom (peaks around -12 dBFS) without pushing quiet passages down into the noise. Deliver at 16-bit or 24-bit as the platform asks. Apple Podcasts, for example, accepts WAV or FLAC at 16 or 24-bit and recommends 24-bit as of October 2026.

SoundStudio itself works in 32-bit floating point while you edit. Float has so much range that a level can go over 0 dBFS inside the editor and come back without damage, as long as it is below 0 when you export. Then you export a 16 or 24-bit WAV. Export writes exactly what is in the editor, so if anything still goes past 0 dB the file clips there, and the message after the export tells you by how much.