Accorcing to the Nyquist theorum, your sampling rate must be greather than twice the highest frequcncy you are trying to sample. Otherwise you lose information at the high frequncy extreme.
2006-12-13 04:39:18
·
answer #1
·
answered by Gene 7
·
1⤊
0⤋
Do answer this precisely, we must know what frequency range speech involves.
Conflicting values are indicated on several web sites I searched, one suggests 100 to 6000 Hz another, up to 9000Hz.
A decent vocal microphone will have a constant frequency response up to arounf 10000Hz, whereas the human ear generally can detect 20 - 20000Hz.
The maximum frequency is of interest, because if you can imagine a sound wave is a sine curve, one full cycle goes from 0 to +1 to 0 to -1 back to 0, so to represent it digitally, you need at a minimum the +1,-1 points - Thus 2 bits per cycle. Of course your sine wave has now become a saw-tooth wave, so the sound reproduction won't be very good - never mind how harmonics effect the wave form.
Thus, if we compromise and assume typical human speech has a max frequency of 8000Hz (cycles per second) we need a minimum of 16Kbs sampling. As explained this would produce poor sound quality, so the 64Kbs standard telephone sampling rate provides 8 bits per cycle instead of the bare minimum 2.
2006-12-13 15:45:42
·
answer #2
·
answered by Leonardo D 3
·
0⤊
0⤋
The standard is 64kbs (DS0) at the ADC (analog to digital converter) conversion, that is what a standard telephone line is.
Pulse Code Modulation PCM rates vary dependent on the device. But that is Sound to Electrical carrier Conversion.
2006-12-13 12:35:40
·
answer #3
·
answered by Anonymous
·
0⤊
0⤋
32kbs is the min. if you want a small file but the sound quality isent that great but 64kbs is the standard for voice
2006-12-13 12:42:05
·
answer #4
·
answered by zjcc5 1
·
0⤊
0⤋