Views: 1553

Written by:

Audio Data Compression: Myths and Controversy Part 1 (Lossy)


This post will focus on lossy audio data compression: the types, the pros and cons, and how to get the most out of the most popular format.

In the audio world you will no doubt run across some very paranoid people who believe that anything that touches your audio will degrade its quality and, in the realm of analog this is often true. One world-class, platinum-caliber mastering engineer that I worked with was convinced that transferring my 24-bit broadcast wav files from my hard drive to his would cause fidelity loss. So, in the interest of easing his concerns, we ran the entire session directly from my drive. Now that digital has firmly seated itself as the storage and transfer medium of choice, it’s important to know what steps do and do not influence the quality of our audio. When something does influence the quality of our audio, it’s important to know when and how. There are, in the most general terms, two kinds of compression: lossy and lossless. Lossy compression reduces the quality of the material while lossless compression can reproduce the original source material perfectly, as if it were never compressed at all. This post is going to focus on the lossy side of things. Tune in tomorrow for a crash course on lossless.

The Lossy Philosophy Explained

The majority of consumers out there deal with lossy compressed audio on a daily basis. In cell phones, (most) DVDs, mp3s, VOIP, lossless is used any time space or bandwidth is of primary concern. The goal of most lossless compression formats is to remove the most data while preserving as much of the perceived quality as possible. In achieving this end, important decisions are made as to where quality can be compromised. As the human voice has a much smaller frequency range than full-range music, cell phones and other voice communication formats will often chop off the high and low end of the frequency range and focus on preserving the critical intelligibility range of 1.5-3k. But for that medium, that choice makes sense. You’ll find a lot of people out there who swear by their music collection of 320kbps mp3s, but I think those people are largely missing the point. Lossy compression is about carefully considered compromises. Simply selecting the highest quality preset is not using lossy compression properly. People who design lossy compression formats do immense amounts of research on what kind of audio information our minds can ignore and what kind of audio information our minds require.

One group of folks that understand this concept quite well are the fellas over at Hydrogen Audio. The folks who created the mp3 standard, Fraunhofer, have long since unofficially passed stewardship of the development to the audio community. And the folks who are on the forefront of this development, painstakingly A/Bing every music codec imaginable, descend upon these forums. These folks are responsible for the LAME mp3 codec which, for the people in-the-know, is the only mp3 encoder worth using.

Community members of HA are also responsible for continued development of the open source Ogg Vorbis compression format. I use aoTuV Beta 5 optimized Vorbis for my personal portable music collection (-Q 5 preset).

They also are developing the open source implementation of the AAC format – the format that Apple prefers for iTunes, iPods and is the ‘official’ successor of mp3.

Hydrogen Audio has done an excellent job at condensing their vast knowledge of digital audio down into their knowledgebase wiki. As far as I am concerned, these three formats: mp3, vorbis, and AAC are the three serious contenders for the best lossy choice. Don’t bother with Windows Media. While the HA wiki does a good job describing the pros and cons of each format, I will run them down briefly here…



  • Widely accepted as THE format. Playable anywhere.
  • Very good quality for bitrate when LAME presets are used.


  • Less efficient than newer codecs.
  • Closed source. Technically people need to pay Fraunhofer for use.
  • Poor efficiency at low bitrates.


  • Open source. No license restrictions for use.
  • More efficient than mp3. Excellent quality at most bitrates.


  • Less supported than mp3. You’ll have to show your friends how to install the Vorbis Quicktime component or buy a hardware player with Vorbis support.
  • What’s a Vorbis? Some people think the name is stupid.


  • More efficient than mp3. Excellent quality at most bitrates.
  • Supported well by Apple and Sony.


  • Less well supported by other major companies and hardware players.

The Best Solution for the Most People

Boiling it down to the best all around solution for distributing music, I use and recommend using LAME 3.97b with the -V 2 preset. Anyone can play it and it should sound transparent.

Should you want to know more, get involved over at Hydrogen Audio and you’ll learn more than you could ever want to know.

Tune in next time for more on lossless compression.

One Response to :
Audio Data Compression: Myths and Controversy Part 1 (Lossy)

Comments are closed.