MP3 and Ogg Vorbis Bitrate Comparison

I had heard enough both good and bad about the sound quality of mp3s to pique my interest. I just ripped Mozart's Don Giovanni (recorded by Giulini) so I wouldn't have to carry it back and forth to lab, so I turned track 17 into mp3s at every supported bitrate. It's not the best recording either technically or for this experiment, but I think that it will show very clearly what's going on, and let you decide what bitrate you can live with.

Guess what? Ogg is getting more popular, as well it should! The recordings here were made using the Vorbis encoder, version 1.0.

Rumour on the street is that bladeenc sucks at low bitrates. That's consistent with these examples! Apparently lame is better, but I don't have time to install it right now.

In the table below, the "bitrate" for the oggs (which use VBR) is the average bitrate. For mp3s it's exact. The files are linked by the flags that I used to do the encoding.

Kb/smp3 (BladeEnc 0.82)ogg (OggEnc v1.0))
32-32
40-40
48-48-b 48
56-56-b 56
64-64-b 64
71-q 1
80-80
83-q 2
96-96
102-q 3
112-112
115-q 4
128-128
137-q 4.99
147-q 5
160-160
176-q 6
192-192-q 7
224-224-q 8
256-256
290-q 9
320-320
434-q 10
1411The uncompressed file (.wav)

For my part, mp3s at 128Kb/s in stereo sounds horrible! 160 is good, 192 is plenty given the quality of my sound card, and 224 is almost indistinguishable from the original. I've been encoding oggs at -q4.99 for my personal use, which is apparently the highest quality that still uses lossy channel coupling, hence the jump between 4.99 and 5 (unusually small in this file, often I've seen it go from 117 k/s to 155 or so). I should note, though, that for the normal soundcard and speakers that came with your Dell, oggenc -q 2 is probably ample!

Ogg Vorbis resources and news