멀티미디어 시스템 – Chepter 2. Multimedia fundamentals

Chapter 2 is about Multimedia fundamentals.

Multimedia fundamentals


Multimedia fundamentals Contents

  1. Informations and Signal
    1. Analog and Digital
  2. Digitization of analog source
    1. ADC and DAC
    2. Bit rate
  3. Digital signal and spectrum
  4. Network

1. Information and Signal

  1. Signal

    1. Physical realization of information in electrical waveform such as current or voltage
      1. Possible to measure in electrical waveform in LAB
      2. Way of carrying out the information
    2. Signal itself carries information
    3. All nature signal (speech, audio, image) exists in the analog form
  2. Information to signal transformation

    • information - Transducer - Signal   * osilloscope
  3. Digital hearing aid System
    digital-hearing-aid-system

    1. Analog signal ( continuous-time signal )
      1. Continuous waveform with respect to time
      2. x(t), y(t)
    2. Digital signal (discrete-time signal)
      1. discrete waveform with respect to time
      2. Sampled version of analog signal x[n], y[n]
      3. 0, 1 is called digital binary
  4. Signal classification

    1. Time-varying : Audio, Spatial-varying : Image
    2. Time+Spatial-varying : Moving Picture = Video
      * Spatial means 2D space
  5. Analog Signal

    1. Information lies on the amplitude
      • x(t) = A ㆍ sin(2πft+θ
      • A - amplitude,  f - frequency in Hz,  θ - phase
    2. Amplitude - strength of the signal
      1. Speech and audio
        1. Represents loudness or strength of the sound
          * SPL ( sound pressure level )
        2. SPL
          1. SPL(dB) = 20 ㆍ log10(P/P0)
            1. P0 is reference sound pressure. It is measured when we can barely hear the 1kHz sine tone.
              P0 = 2.5 X 10-5N/M2
            2. P is measuring sound pressure
      2. Image : Strength or intensity of the light
    3. Frequency (f)
      1. Number of repetitions of same pattern in 1 sec (unit : Hz);
        Period (T) = 1 / f (unit : sec)

        frequency Time-domain waveform Sound hearing
        (a) f=1 relatively slow time-varying relatively heard as low tone (pitch)
        (b) f=5 relatively fast time-varying relatively heard as high tone (pitch)
    4. Phase : Phase angle θ between two sinuoids
  6. Digital Binary

    digital-binary
    - Information lies on discerning ability between binary 0 and binary 1
    - Does not matter with amplitude!!

    1. Advantage of DSP   * DSP : Digital Signal Proccesor
      1. Allow high quality of signal processing
        1. Digital Binary 0, 1 is highly robust to noise
        2. Not sensitive to environmental factors such as temperature as in analog signal
      2. Possible to setup programmable digital system
        1. Can easily change system functionality by slightly modifying the SW program in DSP Chip
        2. Not possible with analog system, the hardware must be redesigned overall
      3. Possible to process multimedia Data
        1. Can easily combine different type of digital media such as speech, audio, image
        2. Easy store, processing, transmission
      4. Low cost digital IC Chip is available  * IC ( integrated circuit )
        1. Getting more smaller and low cost
        2. Low-powered chip is always desirable
      5. Good security messaging services using various coding technique
      6. Weak point of DSP
        1. Need to process huge amount of digital data after A/D conversion
          * So always data compression with digital system
        2. Detection of digital signal require the comms system to be synchronized

 


2. Analog to Digital Conversion

  1. ADC

  2. Two Steps in ADC

    1. Sampling : Sample and Hold
    2. Quantization and Digitization
    3. Digital binary sequence 0110 . . .
      Called - PCM (Pulse Code Modulation)
      resulting binary = PCM code
  3. Sampling

    • Takes samples of analog at regular interval called
    • sampling period
    • sampling rate = =1/ (sample/sec=Hz)
  4. Quantization and Digitization

    1. Qunatization - Process that truncate each sampled value as the ones that computer can represent
    2. Digitization = Process that represent quantized value as digital binary
  5. ADC (Analog to digital conversion)

  6. How Fast one must sample analog signal?

    1. Must fast enough to take consideration of fast varying portion in analog signal.
    2. If sampling speed is too slow, it will lose important high frequency components of the analog signal.
    3. Some trade off in sampling
      Sampling Theory (1950, Shannon)
  7. Nyquist-Shannon sampling theory

    1. Analog signal x(t) can be perfectly reconstructed from its sample values x[n] if we sample analog signal with more than twice the maximum frequency component (or Bandwidth=B) of the analog signal
    2. Nyquist sampling rate =2
      • ex) music signal contains frequency up to 20kHz, what is the Nyquist rate and Nyquist frequency?
        >= (2 * 20kHz = 40kHz)

        • Nyquist rate - 40kHz
        • Nyquist frequency - 20kHz
    3. Aliasing effect and anti-aliasing fliter
      • What if Nyquist sampling condition is not satisfied?
      • No unique analog signal can be reconstructed
  8. Quntization

    1. Trade off in quantization
      • If number of quantization bits (N) is increased
        • High resolution, better representation of sample;
          good reproduction of sound
      • IF N is decreased
        • Signal quality is low, but need to process only a small amount of data
    2. If too few quantization level
      1. Sound : coarse hiss, loss of quiet passages, general fuzziness (quantization noise)
      2. image : banding and posterization
        • banding - dispersing color
        • Posterization - color discontinuity
          posterization1
          posterization2
          posterization3
  9. Bit Rate(Data Rate)

    1. Number of bits to process media signal after ADC to meet signal quality
      Bit rate = R = f (sample/sec) ㆍ N (bits/sample)
    2. Speech signal Bit rate
      1. Analog speech BW ≈ 4KHz
      2. On ADC
        • Sampling rate f = 4KHz ㆍ 2 = 8KHz
          8K = 8000 samples/sec
        • Quantization - 8bit/sample
        • R = 8K ㆍ 8 = 64Kbps
    3. Music signal Bit rate
      1. Music signal BW ≈ 22.05KHz
      2. On ADC
        • Sampling rate f = 22.05KHz ㆍ 2 = 44.10KHz
          44.1K = 44100 samples/sec
        • Quantization - 44.1bit/sample
        • Mono : R = 44.1Kㆍ16 = 0.705Mbps
        • Stereo : Monoㆍ2 = 1.41Mbps

3. Frequency - Domain Spectrum Analysis

  1. Male/female speech discrimination

    1. See frequency spectrum whether it contains high or low frequency component
      • ex 1) DTMT Telephone
        • press 5 - signal composed of 770Hz(low) & 1366Hz(high)
          DTMT Telephone
        • Receiver end – Fourier transform of the signal
          • See frequency spectrum to identify the frequency components
            Receiver end
      • Vuvuzelas noise
        • World cup soccer game 2014, Brazil
        • Vuvuzelas has a constant pitch or frequency of 235 Hz - Use of notch filter
          notch-filter
    2. Two signal representation in DSP
      • Time-domain waveform & Frequency-domain spectrum
        Q) Which one is femal's voice?
        male-femalefemal's voice is seconed image
    3. Bandwidth definition
      bandwidth

      • BW = distance in Hz from 0Hz to max freq. comp.
      • BW is very important concept in DSP and comms area
    4. Revisiting defined of bandwidth
      revisiting-def-bandwidth

      1. What is the BW or max. freq. component in this case ?
        • BW is not the actual max. freq. component
        • Instead, it is frequency composition which has meaningful energy or magnitude values (BW = 4kHz).
      2. Examples of frequency spectrum
        1. Speech word "away" and frequency spectrum
        2. Speech word "away" and frequency spectrum
          speech-ah
        3. 440Hz tuning fork signal
          440khz
      3. Single piano tone (middle C)
        - each piano chord is assigned to single frequency
        piano-tone C
      4. Middle C frequency spectrum
        middle-c

        1. Fundamental frequency (262Hz) + Harmonics
        2. Harmonics are integer multiples of fundamental frequency
      5. Multiple piano chord (CE)
        chord (CE)
      6. Periodic signal + random noise
        random
        From frequency spectrum, easily identify the 32Hz sine wave

4. Networks

  1.  Networks

    • local area networks (LANs) connect several computers on one site (Ethernet)
    • LANs connected together by routers, bridges and switches form an internet.
    • The Internet is a global network of networks (internet) communicating via TCP/IP protocols
      • Mostly operated by commercial Internet Service Providers (ISPs)
  2. Internet Access

    1. Old Dial-up connection uses modem and analog telephone line
      • V90 modem - 56kbps maximum
    2. Broadband always-on digital connection ( > 512kbps)
      • ADSL , Cable, Satellite
    3. Dedicated line (T1, T3)
  3. MIME Types

    1. Need to identify the type of media data in a data stream in a platform-independent way
    2. MIME (Multipurpose Internet Mail Extension)
      1. Originally designed to allow inclusion of data other than text in email, adopted by HTTP
      2. Content-type: type/subtype
      3. Types include text, image, audio, video, application, subtypes define specific formats
      4. .e.g. text/html, image/gif