Dan Pearce

Posted by PH on June 18, 2007
Music & Technology

This is Dan Pearce. He’s just finished doing a BSc (Hons) Music Technology degree at Swansea Metropolitan University. His Major Project was a very high quality piece of work on audio compression codecs.

Perhaps the most interesting part of his project was the listening tests carried out on a set of lossy codecs: MP3, Microsoft WMA, Apple’s AAC, and the open source Ogg Vorbis. Dan conducted a set of tests that compared the subjective quality on each of these based on a set of four criteria: bass response, treble response, clarity, and spaciousness. Each of these in turn was measured against four different types of source material: rock/pop, jazz, classical, and spoken voice. A standard bit rate of 64kbps was used. Here’s a graph showing the compiled results across all the tests:

We have a winner! His results clearly show the superiority of the Ogg Vorbis files in all categories except the classical! WMA and AAC are very closely matched, whilst MP3 consistently performs the worst. Dan puts this down to its age: originally released in 1993 it’s by far the oldest. Here’s a set of samples from Dan’s project which, I think, give a good indication of the relative performances of the codecs:
 
[If you’re interested in more detail here’s the methodology, results, and references from Dan Pearce’s report (5.5MB .pdf). If you want to contact Dan, here’s his email address.]

For me, the findings from this report beg a huge question: why are most people satisfied with the quality of MP3? It’s grainy, harsh, and has a poor stereo image. And yet it would seem that many people are now getting rid of their CD collections and switching entirely to MP3. Yes, of course I can understand the whole slew of benefits afforded by the digital files/downloads/iPod thing, but doesn’t anyone care how bad it all sounds?

And it’s not even as if “CD-quality” audio is all that good. 24-bit/96kHz digital audio is just so much better it’s unreal, and these days it’s pretty affordable too. We have this situation where—for music producers—audio quality has recently shot up, whilst music consumers now seem happy to settle for a substantially lower-quality product than they’ve been used to for the last 20-odd years. Odd, to say the least. Maybe, for most people MP3 is just good enough….

[Note: in order to get the audio examples streaming across the ‘net without glitching I had to subject all the files to a further level of compression. However, they’re all encoded equally at 160kbps stereo, and so the relative differences between them remain the same.]

Tags: , , , , , , , , , , ,

2 Comments to Dan Pearce

  • Why have the samples been recompressed to 160kbps when the original files were all compressed at 64kbps anyway? Won’t this recompression have different effects on the different compression artifacts present with the use of the different codecs? Perhaps mp3 is sufficient because it is just as important to retain maximum compatibility with the rest of the world and at 320kbps mp3 is virtual transparent to ‘CD quality’

  • Wondering where the original files went, it would be far easier to stream those. In my experience, transcoding from 64 to 160 doesn’t really make that much difference if you use a good encoder.
    The reason why people use mp3 still is because it actually does sound pretty good for what it is. Most people don’t use it at 64, they stick to 192 or higher, or a good VBR setting. Lame is getting really good these days.Compatibility is also important, because most consumers probably won’t know what to do with an audio file unless it’s in wav or mp3 format. If they bought a track that they later can’t play, they won’t like it much. But I think most people are finally moving on to newer, better things.

Leave a Reply

Your email address will not be published. Required fields are marked *