Analog to Digital Encoding

Analog data is converted to digital data and then digital signal.

Digitization

Conversion of analog data into digital data.

PCM

Short for Pulse Code Modulation.

Done in 3 steps.

Sampling

Analog signal is measured at regular intervals. Must be sampled at $2f$ frequency where $f$ is the highest signal frequency, as Nyquist theorem says.

Quantization

Each sample is approximated to the nearest value of the finite set of $2^n$ levels. Represented by $n$ bits. The approximation is called quantization error.

If the set of levels are placed linearly, low amplitude signals might get very low precision. To fix that:

• Non-linear quantization can be used. Here, the levels are placed logarithmically. Reduces overall signal distortion.
• Companding can be used, which means compression + expanding. Used to reduce effects of quantization noise especially in low-amplitude parts.

Encoding

Each quantized sample is converted to a binary number.

Delta Modulation

Analog input is approximated by a staircase function. The value is increased or decreased a unit at each sample interval based on the difference between the current sample and the previous sample, which is the delta.