if you sample a signal in regular

intervals of at least twice the highest channel frequency,

the sample will contain enough information to accurately

reconstruct the signal.

Human Audible Range: 20 – 20,000 Hz

Human Speech: 200 – 9000 Hz

Telephone channel: 300 – 3,400 Hz

Nyquist theorem: 300 – 4000 Hz

Sample size = (2 * 4000) 8000, 8000 * 8 bits =

64,000 bps = 64 kbps