Sampling Theorem

The Nyquist-Shannon theorem: a signal can be perfectly reconstructed from samples taken at twice its highest frequency — and what goes wrong (aliasing) if you sample slower.

Sample fast enough and you can reconstruct the signal; sample too slow and a different, lower-frequency wave fits the same dots (aliasing)
Sampled above Nyquist rate — reconstructableSampled below Nyquist rate — aliasedsame dots, two very different waves
Definition

To store or process a continuous signal (like sound) digitally, you must sample it — record its value at regular instants in time. The Nyquist-Shannon sampling theorem says: a signal can be perfectly reconstructed from its samples if and only if it was sampled at a rate of at least twice its highest frequency component.

That minimum rate, 2fmax2f_{\max}, is called the Nyquist rate. Sample any slower, and information is permanently lost — different signals become indistinguishable from their samples, a phenomenon called aliasing.

Why CD audio is sampled at 44,100 Hz

Human hearing tops out around 20,000 Hz. To safely capture all audible frequencies, CD audio samples at 44,100 Hz — comfortably above the Nyquist rate of 2×20,000=40,0002 \times 20{,}000 = 40{,}000 Hz.

Try it

If the highest frequency in a signal is 5,000 Hz, what is the minimum sampling rate needed to reconstruct it without aliasing?

Solution

10,000 Hz (10 kHz) — twice the highest frequency, per the Nyquist rate.

Related concepts