VoIP Foundations: How Spoken Voice Becomes a Packet

1. Sample the analog signal a lot (8000 times a second actually)

2. Create a number that represents each sample (Aka Quantization)

3. Convert that number to binary (PCM Scale: -127 TO +127 – MULAW/ALAW)

4. Compress if desired

‘Nyquist Therom’

Human voice frequency range – 20-9000hz

20-4000hz range = ‘meat’ of human voice

Sample the highest core frequency by 2 per second = 8000

The reason for low quality voice on AM radio is to only the ‘meat’ of human voice.

Goal: Recognise the voice and tell their mood!

PCM Scale – ‘Post Code Modulation’

8000 times a second it will state the binary number

Non-linear scale

127 up and 127 down on Left Axis (256 is highest number from 8 bits)
Time in 1 second intervals via Right Axis

8000 x 8 = 64Kbps

ulaw and alaw with G.711.. (PCM Scale)
ulaw is USA and is backwards to alaw. alaw just makes sense!