多媒体通信

授课教师：黄晓燕

DM(Delta Modulation)

simplified version of DPCM. Uniform-DeltaDM:use only a single quantized error value. ——>1-bit coder.

if e_n>0 ,e_k=+k; else e_k=-k;

Adaptive DM:change the step size adaptively.

ADPCM(adaptive DPCM)

Making the predictor coefficients adaptive

Contents

MPEG Audio

Psychoacoustic（心理声学） The range of human hearing is about 20Hz to 20KHz.

The frequency range of the voice is typically only from about 500Hz to 4KHz

Threshold of Hearing

Fletcher-Munson Equal loudness Curves

Human ear's sensitivity is different for different frequency.Most sensitive:1KHz - 5KHz

Frequency masking（频率掩蔽）

if a very loud tone is produced , it is impossible to hear any sound nearby in the frequency spectrum.

纯音的掩蔽基本符合以下几个规律：低音容易掩蔽高音，高音较难掩蔽低音；频率相近的纯音容易互相掩蔽；提高掩蔽声的声压级时，掩蔽阈会提高，而且被掩蔽的频率范围会扩展。

Critical bands（临界频段）

Represents the ear's resolving power for simultaneous tones or partials.

Because of masking , the ear is not very discriminating within a critical band.

At the low frequency, the band is small, when the frequency get higher, the band becomes wider.

Temporal Masking （时间掩蔽）

Test tones with frequencies near the masking tone are the most masked.

MEPG Audio (Moving Pictures Experts Group)

Most of the complexity increase is at the encoder,not the decoder.