PROBLEMS WITH ANALOG CONNECTIONS
- Distance limitations – analog signals eventually fade. Repeaters can be introduced to regenerate signal but also regenerate anything else picked up including white noise.
- Wiring requirements – for each call, a tip and ring wire is required. Not scalable.
Digital signals convert analog wave into zeroes and ones.
DIGITIZING VOICE
Step 1 – Sample the signal
- Nyquist formula: If you can sample at twice the highest frequency, you can accurately reconstruct a signal digitally.
- Common frequencies – Human ear: 20 – 20,000Hz. Human speech: 200 – 9000Hz
- Nyquist theorum: 300 – 4000Hz.
Step 2 – Perform Quantization on the Sample
- Lining samples up to a scale that the samples can be represented with. In this case taking the value of the amplitude and line it up to a scale (known as Pulse Amplitude Modulation (PAM). The Y access on the chart are labelled segments. More samples taken around the baseline.
Step 3 – Convert the “quantized” signal to binary
- Known as pulse code modulation (PCM).
- 2 PCM methods: A-LAW and MU-LAW.
- A-LAW makes more sense.
- Mu-Law used in US, Japan and Canada
- A-LAW used elsewhere. First digit: 1 presents positive numbers . 0 represents negative numbers. Next three digits represent Segment number. Next four digits are interval.
- Mu-Law – The exact opposite of A-LAW.
Step 4 – Optionally compress the samples
You can
- Send all these samples
- Send just the changes
- Build a code book
Standard voice sample is 64Kbps. Common compressed value: 8kbps (G.729 codec)
