This document discusses sampling theory and how it relates to digitizing audio signals. It covers:
1) How analog signals can be represented in both the time and frequency domains.
2) The Nyquist sampling theorem which states that a signal must be sampled at least twice the maximum frequency to avoid aliasing.
3) Key parameters for digitizing audio like sampling rate, bit depth, and standards like CD quality 44.1kHz sampling.
4) Sources of error and limitations in converting digital audio back to analog signals like filter imperfections, quantization error, and clock jitter.
Copyright:
Attribution Non-Commercial (BY-NC)
Available Formats
Download as PPT, PDF, TXT or read online from Scribd
This document discusses sampling theory and how it relates to digitizing audio signals. It covers:
1) How analog signals can be represented in both the time and frequency domains.
2) The Nyquist sampling theorem which states that a signal must be sampled at least twice the maximum frequency to avoid aliasing.
3) Key parameters for digitizing audio like sampling rate, bit depth, and standards like CD quality 44.1kHz sampling.
4) Sources of error and limitations in converting digital audio back to analog signals like filter imperfections, quantization error, and clock jitter.
This document discusses sampling theory and how it relates to digitizing audio signals. It covers:
1) How analog signals can be represented in both the time and frequency domains.
2) The Nyquist sampling theorem which states that a signal must be sampled at least twice the maximum frequency to avoid aliasing.
3) Key parameters for digitizing audio like sampling rate, bit depth, and standards like CD quality 44.1kHz sampling.
4) Sources of error and limitations in converting digital audio back to analog signals like filter imperfections, quantization error, and clock jitter.
Copyright:
Attribution Non-Commercial (BY-NC)
Available Formats
Download as PPT, PDF, TXT or read online from Scribd
This document discusses sampling theory and how it relates to digitizing audio signals. It covers:
1) How analog signals can be represented in both the time and frequency domains.
2) The Nyquist sampling theorem which states that a signal must be sampled at least twice the maximum frequency to avoid aliasing.
3) Key parameters for digitizing audio like sampling rate, bit depth, and standards like CD quality 44.1kHz sampling.
4) Sources of error and limitations in converting digital audio back to analog signals like filter imperfections, quantization error, and clock jitter.
Copyright:
Attribution Non-Commercial (BY-NC)
Available Formats
Download as PPT, PDF, TXT or read online from Scribd
Download as ppt, pdf, or txt
You are on page 1/ 34
Sampling Theory
Time domain • Present a recurring phenomena as amplitude vs. time Sine Wave – Sine Wave 1.5 1 Amplitude
0.5 0 -0.5 0 1 2 3 4 5 6 -1 -1.5 Time Frequency domain • Present recurring phenomena as amplitude vs. frequency • Same sine wave looks like – Amplitude
Frequency Multiple Waves Multiple Waves Both Domains Voice in both Domains
Voice in the Time Domain
Voice in the Frequency Domain Harmonics • See Spreadsheet
Flute Clarinet
Horn Guitar Fourier Analysis
• The eardrum responds to a sum of all the
waves arriving at a particular instant. Yet the individual sounds are “heard.”
• Any waveform is composed of an infinite
number of simple sine waves of various frequencies and amplitudes. Diatonic C Major Scale Letter Frequency Frequency Note Name (Hz) ratio Interval do C 264 9/8 Whole re D 297 10/9 Whole mi E 330 16/15 Half fa F 352 9/8 Whole sol G 396 10/9 Whole la A 440 9/8 Whole ti B 495 16/15 Half do C 528 Poor Sampling 1.5
0.5
0 0 2 4 6 8 10 12 -0.5
-1
-1.5
Sampling Frequency = 1/2 X Wave Frequency
Even Worse 1.5
0.5
0 0 2 4 6 8 10 12 -0.5
-1
-1.5
Sampling Frequency = 1/3 X Wave Frequency
Higher Sampling Frequency 1.5
0.5
0 0 2 4 6 8 10 12 -0.5
-1
-1.5
Sampling Frequency = 2/3 Wave Frequency
Getting Better 1.5
0.5
0 0 2 4 6 8 10 12 -0.5
-1
-1.5
Sampling Frequency = Wave Frequency
Good Sampling 1.5
0.5
0 0 2 4 6 8 10 12
-0.5
-1
-1.5
Sampling Frequency = 2 X Wave
Frequency Shannon-Nyquist's Sampling Theorem • A sampled time signal must not contain components at frequencies above half the sampling rate (The so-called Nyquist frequency)
• The highest frequency which can be
accurately represented is one-half of the sampling rate Range of Human Hearing • 20 – 20,000 Hz • We lose high frequency response with age • Women generally have better response than men • To reproduce 20 kHz requires a sampling rate of 40 kHz – Below the Nyquist frequency we introduce aliasing Effect of Aliasing • Fourier Theorem states that any waveform can be reproduced by sine waves. • Improperly sampled signals will have other sine wave components. • http://www2.egr.uh.edu/~glover/applets/Sampling/Sampling.html Half the Nyquist Frequency 1.5
0.5
0 0 5 10 15 20 25
-0.5
-1
-1.5 Nyquist Frequency 1.5
0.5
0 0 2 4 6 8 10 12
-0.5
-1
-1.5 Digitizing Key Parameters • Sampling frequency – 11.025kHZ or 22.05kHZ or 44.1kHZ • Number of bits per sample – 8 bits (256 levels) or 16 bits (65,536 levels) Digital Voice Telephone Transmission • Voice data for telephony purposes is limited to frequencies less than 4,000 Hz. • According to Nyquist, it would take 8,000 samples (2 times 4,000) to capture a 4,000 Hz signal perfectly. • Generally, one byte is recorded per sample (256 levels). One byte is eight bits of binary data. • (8 bits * 8,000 samples per second = 64K bps) over a circuit. T-1 Transmisson • T carrier circuits are designed around this requirement, since they are primarily designed to carry analog voice signals that have been digitalized. • For example, look at the DS-1 signal which passes over a T-1 circuit. For DS-1 transmissions, each frame contains 8 bits per channel and there are 24 channels. Also, 1 "framing bit" is required for each of the 24 channel frames. T-1 Transmissons • (24 channels * 8 bits per channel) + 1 framing bit = 193 bits per frame. 193 bits per frame * 8,000 "Nyquist" samples = 1,544,000 bits per second. • And it just so happens that the T-1 circuit is 1.544 Mbps.--not a coincidence. Each of the 24 channels in a T-1 circuit carries 64Kbps. Standards • DS0 - 64 kilobits per second • ISDN - Two DS0 lines plus signaling (16 kilobytes per second), or 128 kilobits per second • T1 - 1.544 megabits per second (24 DS0 lines) • T3 - 43.232 megabits per second (28 T1s) • OC3 - 155 megabits per second (84 T1s) • OC12 - 622 megabits per second (4 OC3s) • OC48 - 2.5 gigabits per seconds (4 OC12s) • OC192 - 9.6 gigabits per second (4 OC48s) Quantization Error
Approximation or quantizing error
Greater error = more noise
D/A Conversion CD ROMS • Sampling rate is 44.1 kHz • Nyquist Theorem says that the highest reproduced frequency is 22.05 kHz. – Any frequency above 22.05 kHz will produce aliasing • A low pass filter is used to block frequencies above 22.05 kHz. Problems with D/A • Imperfect low pass filters • Ideally you want 0 dB attenuation at 20 kHz going up to 90 dB at 22 kHz – Very expensive • Oversampling will help – Sample at 8 X 20 kHz = 160 kHz • Then the low pass filtering needs to be accomplished in 140 kHz not 2 kHz Problems with D/A • Finite word length – Most systems today do 16 bit digitizing – 65536 different levels • The loudest sounds need room, so the normal sounds don’t use the entire range – Problems occur at the low levels where sounds are represented by only one or two bits. High distortions result. • Dithering adds low level broadband noise Problems with D/A • Clock speed variation (Jitter)
