MP3 - The Music Standard
MP3 stands for Motion Picture Experts Group (MPEG) 1 layer 3.MPEG is an audio-video standard setting body. This group works on standards for the coding of moving pictures and audio. MPEG-1 layer 3 is a specification for storing audio information in a compressed format (say zipped).
A negligible amount of compression is achieved if we use PKZIP to compress multimedia data. This is because programs like PKZIP look for repeating patterns of bits and bytes, and replace them with tokens to reduce file size. Although multimedia files, particularly music, contain repeating patterns, they are not apparent after digitization. The bits and bytes that correspond to two different sections of music will not be identical even if they sound the same to the human ear. So to compress an audio file effectively one needs to exploit the way humans perceive sound. Such encoding schemes are called perceptual encoding schemes.
The bit rate is the amount of compression required, usually expressed in kilobytes per second (KBPS). Low bit rate means high compression.
Gains and losses:
MP3 is a perceptual encoding scheme that compresses audio files to about 1/12th of its original size with a negligible loss of quality. There is however, one major point of distinction between a zip file and a MP3 file. When you expand a zip file you get back an exact copy of the original file with no loss of information. This is called lossless compression. But decompressing an MP3 file will not give you exact copy of the original source file. Information is irretrievably lost during MP3 encoding process.
Hardware and Software:
To playback MP3 files, the recommended minimum specification is a 486DX4 100MHz machine. To create your own MP3 files, you need a powerful PC- 200 Pentium or above. You need a sound card.
If you are planning to create MP3 file from audio, better use batteries than mains (because noise is less). Your sound card should be able to sample at 44100 samples per second (or 44.1 kHz).
If you are planning to create MP3 files from music CD s, you should preferably have a sound card that accepts digital inputs and a CD-ROM drive that allows for DAE (Digital Audio Extraction). If these are not present, it is still possible to create MP3 files, though at slightly reduced quality.
If takes 1.5 GB (Approx.) to hold wave files, and the temporary files, to create a 30 to 45 minute MP3.
Capturing Audio:
Once you have connected the music system to the sound card, you need a program to capture the sound coming into your sound card. Windows comes wi
th an applet called sound recorder, which however records to system RAM. When it runs out of RAM, it quits. It is much better to have software that records to the hard disk ex: cool Edit 96 from syntrillium software corporation.
If you plan to create MP3 from music CD s, use audiograbber software that creates a wave file from audio CD s is called ripping.
Removing Noise:
The captured wave file needs to be cleared up before it can be compressed to MP3.
CLICKS, TICKS AND POPS: The wave format includes a header and optionally footer with the music contained within. To remove clicks and ticks at the beginning where there is no music, simply zoom into those areas, select the portions that contain the clicks, ticks etc and simply delete them. Alternatively, you can replace them with silence.
TAPE HISS AND OTHER FIXED TONES: Tape hisses is usual when you record from an audiocassette. This is one area where cool edit really excels. First select an area of the wave which contains only this type of noise. This section should be atleast 0.5 seconds. After selecting the area with the noise (and without the signal), TRANSFORM --> NOISE REDUCTION. You get a dialogue box containing a button titled get noise profile from selection. Upon clicking this button, cool Edit will generate a noise profile based on the selection. A noise profile enables cool Edit to distinguish between noise and signal. Now zoom out and select the entire wave file. With the whole wave file selected use TRANSFORM>NOISE REDUCTION. This operation takes about one and half hour on 450 MHz Pentium III.
The noise reduction to 80% is acceptable. To reduce noise to 100% may result in distortion of original signal.
Compression:
The final step is to encode the wave to MP3. The MP3 standard does include an algorithm written in pseudo-code for demonstration purposes, but the actual implementation of the encoding process is left to the developers.
The best encoder is L3ENC. This encoder has a command line interface that offers powerful features for those used to graphical interface. Another popular encoder is offered by Xing technologies. There is also a freeware-encoding engine, called blade.
The speed of compression is one of the important factors. To compress a 45-minute wave file L3ENC will take about 2 hrs. Others take less time but the quality is usually inferior.
Web links: The links below should take you to the software mentioned in the article. In case the pages have moved, try searching for them at a popular search engine.
SYNTRILLIUM SOFTWARE CORPORATION:
www.syntrillium.com AUDIOGRABBER:
www.audiograbber.com-us.net FRAUNHOFER INSTITUTE:
www.iis.fhg.de/audio