**Nyquist–Shannon Sampling Theorem**

E.g. 20,000 Hertz into this formula will not be reconstructed perfectly at anything less than 0.00003 seconds per sample. Recording at a sample rate of 48kHz is sampling at 0.00002 seconds per sample, hence for situations like recording dialogue where the highest frequency for normal conversation is below 20kHz (because we can’t hear higher anyway).