The analog signal m(t) with the highest frequency less than  can be uniquely determined by its equally spaced sampling values. The sampling interval Ts or the sampling rate fs should satisfy:

Waveform and spectrum of ideal sampling process: fs≥2fH

Aliasing distortion: fs≤2fH

The original signal cannot be reconstructed without distortion, and the sampling rate must meet:

Sampling and recovery principle block diagram:

The relationship between fs and fL

Naturally sampled PAM

PAM with flat top sampling

Represent infinite sample values ​​with a finite number of quantization levels

Divide the value range of the input signal at equal intervals

Signal to noise ratio S/Nq

Disadvantages of Uniform Quantization

application

Quantization method with unequal quantization intervals

Compression-expansion characteristics:

Two recommendations from the ITU

PCM system schematic block diagram

Analog signal digitization process---"sampling, quantization and encoding"

Commonly used binary codes

One of the issues to consider when coding

Code bit selection and arrangement

Start level and quantization interval

DPCM principle

DPCM performance

Data compression does not allow any loss, so only lossless compression methods can be used.

This requires choosing an efficient encoding to represent the source data to reduce the redundancy of the source data.

Since the information content of each character in the finite discrete source is different, variable length codes are usually used for compression.

In order to determine the demarcation of each character of the variable length code, a unique decodability is required.

Huffman code is a commonly used variable-length code without prefix, which is the optimal code in the sense of minimum code length.

Huffman code encoding process

Compression encoding performance indicators

Category 1

Category 2

Still image compression exploits correlations between neighboring pixels and often performs lossy compression in the transform domain. The most widely used international standard for still image compression is JPEG.

Dynamic image compression utilizes the correlation between pixels in adjacent frames, and then tries to reduce the correlation between pixels in adjacent frames on the basis of still image compression. The most widely used international standard for dynamic image compression is MPEG.

waveform encoding

Parameter encoding

mixed encoding

Articulatory organs and principles of pronunciation

Speech parameters and their extraction methods

Disadvantages of parametric encoding

Ways to improve

At the encoding end, the n-dimensional input signal vector x is compared with each codeword in the code book to find the codeword qi with the smallest distortion; then its number i (encoded) is transmitted to the decoding end. At the decoding end, after receiving i (encoding), the value of i is obtained through decoding, and then found from the code book. Obviously, vector quantization is a lossy compression encoding, but its compression performance is better than scalar quantization. . Find the quantized vector qi of x.

(a) Time division multiplexing principle

E architecture diagram:

Frame structure of PCM primary group:

Incremental modulation principle block diagram

Delta modulation waveform diagram

Maximum tracking slope of the decoder

No overload conditions:

Coding range:

Signal maximum power: