voice signal sampling

[FERMED]

Hi there,

Could someone tell how 11025Hz, 22050Hz and 44100Hz are chosen to sample voice signal? what I mean is why exacly this values?

For example for 11025Hz, why this value and not 11000Hz or 12000Hz?
Where does all this come from?

Many thanks

 

[fcfusion]

The human ear can detect sounds up to 20 kHz. In order to represent the sound you need to sample it at twice the maximum frequency (40 kHz) at least. This is the Nyquist theorem and you probably already heard about it.

In theory, if you respect the Nyquist theorem you can recover your signal by using an ideal low-pass filter. However such filters do not exist so you have to sample your signal with a little higher frequency so you can use a non-ideal filter, in this case 44100 Hz is standard frequency, considered to be enough.

For lower frequencies same thing is applied.