In this web page, we introduce

• the basic facts about sound,
• how sound are digitized using PCM coding
• popular MP3 information
• recording voice/music using Cool Edit 96.

Reference: Material here adapted from the following Related web pages and literature.

• Davis Pan, "A Tutorial on MPEG/Audio Compression", IEEE Multimedia, pp. 60-74, 1995.
• Audio Compression Overview at Simon Fraser University

A good technical overview of psychoacoustics, MPEG Audio and other audio compression algorithms.
• Tutorial paper on MPEG/Audio Compression by Davis Pan
• Dolby AC-3: Multichannel Perceptual Coding at Dolby
• FAQ about MP3, MPEG Audio Layer 3 by Fraunhofer Institute.
• Modern Audio Technology by Martin Clifford, 1992, Prentice Hall,
• Data and Computer Communications, by William Stallings, 5th edition, 1999, Macmillan.

• Longitudinal wave: medium vibrate in the same direction of wave advance.
• Sound can be converted via transducer to other form of energy for storage.
• Inverse Square Law: the intensity of the sound radiation decreases in proportion to the square of its distance from the sound source.
• Acoustic masking- one sound can cover (mask) another making it inaudible (especially in mid- to treble range)
• Sound Pressure Level (SPL) is a variation above and below normal atmospheric pressure.
• Sound depends on the production of changes in atmospheric pressure.
• Audio band- 20-20kHz

0-20Hz- infrasonic
above 20kHz- ultrasonic (supersonic)

Hearing-response Characteristics

Basic characteristics of a sound wave

Harmonics and Tone Color

The same note play on different musical instruments generate different tone.
The harmonics composed in tones make them different.

Digital vs. Analog

Reality: signal got distorted over distance because

1. impedance (within transmission medium)
2. interference (outside force, e.g. cloud, lightening)

Digital signal can be regenerated better without the distortion experienced in analog signal.

Recording analog audio signal using Pulse Code Modulation (PCM)

Analog signal from microphone or telephone handset are sampled at high frequency, for telephone it is 8000 samples/sec, for digial music recoding, it is 44.1 samples/sec.  Each sample is then represented by 8 bits or 16 bits resprectively.  The resulting is a digital bit stream.

Quantization error can be reduced by having higher sampling rate (sample more frequently)and more quantization level (more bits for each sample).

Example: With Stereo, 16 bits/sample, 44kHz sampling rate, PCM encoding, how many bits of data will be generated by a three minute sound recording?
Ans: 3min*60sec/min*44000samples/sec*16bits/sample*2(channel)/8bits/byte=31.68MB/s

Masking

• A phenomenon of the human hearing system. Normal human ears are sensitive to a wide range of frequencies. However, when a lot of signal energy is present at one frequency, the ear cannot hear lower energy at nearby frequencies.
• We say that the louder frequency masks the softer frequencies. The louder frequency is called the masker.

• The basic idea of SBC is to save signal bandwidth by throwing away information about frequencies which are masked.
• The result won't be the same as the original signal, but if the computation is done right, human ears can't hear the difference.
• Divide signal into bands and perform masking computation and throw away weak signals in each band.

• It is a perceptual audio coding scheme, exploiting the masking property of the human ear, and trying to maintain the original sound quality as far as possible.
• MPEG Audio specifies a family of  three audio coding schemes, simply called Layer-1, Layer-2, and Layer-3. From Layer-1 to Layer-3, encoder complexity and performance (sound quality per bitrate) are increasing.
• In MPEG 1 Audio standard,
• all three layers may use 32, 44.1 or 48 kHz sampling frequency.
• All Layers are allowed to work with similar bitrates:


                             Layer-1: from 32 kbps to 448 kbps
                             Layer-2: from 32 kbps to 384 kbps
                             Layer-3: from 32 kbps to 320 kbps

• In MPEG-1: max. 1.5 Mbits/sec for audio and video, About 1.2 Mbits/sec for video, 0.3 Mbits/sec for audio
• In MPEG 2 Audio standard, there two new extensions:
• low sample rate extension: extend sampling rates to 8, 16, 22.05 or 24 kHz.
• multichannel extension: address surround sound applications, with up to 5 main audio channels (left, center, right, left surround, right surround) and optionally 1 extra "low frequency enhancement (LFE)" channel for subwoofer signals;
• multilingual extension: allow the inclusion of up to 7 more audio channels.
• Compression ratios: MP3 achieve 1:10-1:12 compression ratio, at about 64 kbit/s per audio channel, while maintaining the original CD sound quality .
• MPEG Audio Algorithm

Steps in algorithm:

1. Use convolution filters to divide the audio signal (e.g., 48 kHz sound) into frequency subbands that approximate the 32 critical bands --> sub-band filtering.
2. Determine amount of masking for each band caused by nearby band using the results shown above (this is called the psychoacoustic model).
3. If the power in a band is below the masking threshold, don't encode it.
4. Otherwise, determine number of bits needed to represent the coefficient such that noise introduced by quantization is below the masking effect (Recall that 1 bit of quantization introduces about 6 dB of noise).
5. Format bitstream

 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 



Example:

• After analysis, the first levels of 16 of the 32 bands are these:

----------------------------------------------------------------------
Band        1  2   3   4  5  6   7   8   9  10  11  12  13  14  15  16  
Level (db)  0  8  12  10  6  2  10  60  35  20  15   2   3   5   3   1
----------------------------------------------------------------------

• If the level of the 8th band is 60dB,

 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

it gives a masking of 12 dB in the 7th band, 15dB in the 9th.

Level in 7th band is 10 dB ( < 12 dB ), so ignore it.

Level in 9th band is 35 dB ( > 15 dB ), so send it.

--> Can encode with up to 2 bits (= 12 dB) of quantization error.

MPEG Layers

• MPEG defines 3 layers for audio. Basic model is same, but codec complexity increases with each layer.
• Divides data into frames, each of them contains 384 samples, 12 samples from each of the 32 filtered subbands as shown below.

 

 
 



















Figure: Grouping of Subband Samples for Layer 1, 2, and 3

• Layer 1: DCT type filter with one frame and equal frequency spread per band. Psychoacoustic model only uses frequency masking.
• Layer 2: Use three frames in filter (before, current, next, a total of 1152 samples). This models a little bit of the temporal masking.
• Layer 3: Better critical band filter is used (non-equal frequencies), psychoacoustic model includes temporal masking effects, takes into account stereo redundancy, and uses Huffman coder.

• CoolEdit 2000 is a shareware that records, playbacks, and editing sound files. (GEt 30 day trial version with limited features on http://www.syntrillium.com/cooledit/).
• Select Start | Program | Cool Edit 2000 | Cool Edit 2000 and the following main window appear.


• Select File | "new" menuitem, "New Waveform" dialog window appear.
  
• Specify the sampling rate, resolution (bits per sample), and channels(mono/stereo).


For human speech we can get by with 11kHz, 8bit, and mono setting. 
For singing, we can increase from 11 kHz to 22 kHz.
For music recording, use 44kHz and 16 bit.

Hit OK after choosing the sampling parameters.
• Press red circle "Record" button on the lower left corner of the screen to start recording.

Hit square "Stop" button when you are done.
The recorded wave form will be displayed on  the main window.

• If the signal is weak,
  Select Transform | Amplify | Normalized menuitem  to enhance the signal to a "normal" level.
• If did not get any recorded signal, make sure the microphone volume is reasonable high and the mute control is not checked.
  
• Editng the sound file by  selecting the area (point and drag) of the wave form and hit control-X or Edit | Cut menuitem.
• Select File | save as menuitem.  Choose filename and audio file type, typically   .au (Sun/Apple), .ra (RealNetworks), .mp3 and  .wav (Microsoft) formats.

Here are three sample audio files generated by CoolEdit, click to play them:
Vincent singing 179KB, au format and
Vincent singing, 33KB, realaudio format and
Vincent singing, 182KB, wav format.

Homework Exercises:

1. With mono, 8 bits, 22kHz, PCM encoding, how many bytes of data will be generated by a three minute sound recording?
2. Create audio and link to a web page.
• Use CoolEdit to record a less than 5 second voice with mono, 16 bit, 22kHz, encoding.
• Edit out the unnecessary silence portion of the sound track and apply the normalized special effect.
• Save as both .wav and .ra file formats with your_login.wav and your_login.ra as file names.
• Ftp your audio files to your public_html/sounds directory.
• Create hyperlinks in your class personal web page to the audio files


using  <a> tag such as
<a href="sounds/your_login.ra">Vincent singing, 33KB, realaudio format</a>

• Indicate the file size for comparison purpose.