Sampling theorem: The idea
In plain words
The idea explained without jargon
A sound or radio wave is a smooth curve. If you measure its height at regular instants and write down the numbers, you might expect to lose whatever happens between the measurements. The sampling theorem says you lose nothing, as long as the signal contains no frequencies above some limit and you measure at least twice per second for every hertz of that limit [1]. Then only one smooth curve passes through your numbers, so the original can be rebuilt exactly. Measure too slowly and fast wobbles are mistaken for slow ones, which is called aliasing. Real signals never stop exactly at a limit, so engineers filter them first and sample a little faster than the bare minimum [6].
as of 2026-11-12
Key equation
The equation that sums it up, and what each symbol means
- Equation
- f_s >= 2 W
- Symbols
- {"f_s":"the sampling rate, in samples per second","W":"the highest frequency in the signal, in hertz"}
- Note
- Shannon wrote it as samples spaced 1/(2W) seconds apart, that is 2W samples a second [1]. Rebuilding the signal means adding together one smooth 'sinc' pulse per sample. A component lying exactly at W can be missed at exactly 2W, so in practice the rate is kept somewhat above 2W [6].
as of 2026-11-12
An analogy
A picture from everyday life that makes the idea easier to hold
Think of a film of a spinning wagon wheel. A film is a set of still pictures taken at a fixed rate. If the wheel turns slowly, the pictures show its motion truthfully. If it turns too fast for the frame rate, it seems to creep or even spin backwards. The picture stops being true because a wheel has one speed, while a real signal mixes many frequencies, and the theorem needs every one of them to stay below the limit.
as of 2026-11-12
Sources
- Communication in the Presence of Noise (C. E. Shannon; Proc. IRE 37(1), 10-21, 1949; Classic Paper reprint, Proc. IEEE 86(2), 1998) Proceedings of the IEEE (reprint, copy hosted by the University of Miami), 1949
- Explanation of 44.1 kHz CD sampling rate Columbia University, Department of Computer Science