Haggs
AI Team Member Joined: 05 May 2004 Posts: 2723 Location: Minneapolis, Minnesota	Posted: Fri Oct 08, 2004 2:13 pm    Post subject: The Ingenious MP3 Compression Algorithm   Over the past 5 years MPEG Audio Layer 3 files, more commonly known as MP3s, have flourished due to a very creative and intelligent compression algorithms. While literally millions, if not billions, of people have hundreds, if not thousands, of these MP3 files on their computers, few probably understand and appreciate what a gem the MP3 really is. Prior to the development of MP3 technology, music was typically found in WAV file format which has a ten times greater file size. MP3 listeners know that there is virtually no difference in noticeably audio quality between listening to CDs directly versus listening to an MP3 encoded at 128kb/s (kilobytes per second). How could a file be compressed ten fold without hardly any loss in audio quality? MP3 files use "lossy compression" with "psychoacoustic algorithms". Lossy compression is defined as compression that occurs from taking away bits of information. In terms of an MP3, this means that an MP3 ripper removes audio information from the original WAV file. However, this lossy compression is done with psychoacoustic algorithms. Psychoacoustic algorithms are methods that describe how to selectively remove audio information that the human ear would most likely be unable to notice anyways. For example, the psychoacoustic algorithms usually leave the 1 KHz to 4 KHz audio frequencies untouched, because this is the frequency range that the average human ear can hear the best. The algorithms also remove quieter audio sounds that are playing at the same time as louder ones, since the quieter audio sounds would not likely be noticed present or not. The third way that the algorithms are able to compress the file is by converting stereo signals into mono signals. Deep bass is difficult for the human ear to detect direction, so it is very difficult for the average ear to tell whether the bass sounds are coming from the right or the left, or mono. Since mono audio requires fewer bits of information, the file is compressed further. Last, the file size is further reduced by optimizing the data for the most frequently occurring signals. So next time you queue up your play list in Winamp, iTunes, or your favorite MP3 player, you can appreciate the intellectual genius behind each track.
Back to top