Skip to main content
Topic: AAC vs. Vorbis (Read 2377 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

AAC vs. Vorbis

I did an ABX recently with the killer sample castanets and hi-hat file.   I was very surprised that Vorbis, which I have not used in some time (I've been muddling along with LAME for a while, being out of the loop for the past couple of years), outperformed Apple's AAC codec.  I use Aoyumi's latest tunings.  Both Opus and Vorbis were equally good at 128 kbps, even testing using planarmagnetic and electret headphones.  But the ACC was relatively easy to spot 6-7 out of 8 times, repeatedly.

I'm 42 and I only hear out to 16 KHz, so my hearing is not perfect.  Perhaps Q4 Vorbis is not transparent to everyone?

Re: AAC vs. Vorbis

Reply #1
[...] I only hear out to 16 KHz, [...]
try tonegenerator, like SineGen, to make sure & calibrate sound complex (ears+HW)
q4/128kbps transparent... never! q8/256 minimum.

Re: AAC vs. Vorbis

Reply #2
[...] I only hear out to 16 KHz, [...]
try tonegenerator, like SineGen, to make sure & calibrate sound complex (ears+HW)
q4/128kbps transparent... never! q8/256 minimum.

I have heard 16 KHz on Bose speakers, and I can hear it on my setup using Koss UR40's, though it is extremely quiet, almost inaudible at normal listening volumes.  I borrowed Oppo headphones to do the test, and I also used my own Electrostatz to confirm the result, because I wanted a more analytical sound.

I don't know what kind of artifacts one would be hearing at 128kbps with Vorbis that one would not hear also at 256kbps.  The pre-echo of Vorbis seems to have been greatly improved since I tried it years ago.   Of course, maybe my hearing is also worse.

Re: AAC vs. Vorbis

Reply #3
I did an ABX recently with the killer sample castanets and hi-hat file.   I was very surprised that Vorbis, which I have not used in some time (I've been muddling along with LAME for a while, being out of the loop for the past couple of years), outperformed Apple's AAC codec.  I use Aoyumi's latest tunings.  Both Opus and Vorbis were equally good at 128 kbps, even testing using planarmagnetic and electret headphones.  But the ACC was relatively easy to spot 6-7 out of 8 times, repeatedly.

I'm 42 and I only hear out to 16 KHz, so my hearing is not perfect.  Perhaps Q4 Vorbis is not transparent to everyone?

Just a heads up. fdkaac is much smaller than Apple's codec and the quality is roughly the same, so instead of shoehorning the worst Windows program ever onto your PC (iTunes), or even worse for Linux users, figuring out how to get it to work on Wine, you could just use fdkaac.

Re: AAC vs. Vorbis

Reply #4
[...] I only hear out to 16 KHz, [...]
try tonegenerator, like SineGen, to make sure & calibrate sound complex (ears+HW)
q4/128kbps transparent... never! q8/256 minimum.

I have heard 16 KHz on Bose speakers, and I can hear it on my setup using Koss UR40's, though it is extremely quiet, almost inaudible at normal listening volumes.  I borrowed Oppo headphones to do the test, and I also used my own Electrostatz to confirm the result, because I wanted a more analytical sound.

I don't know what kind of artifacts one would be hearing at 128kbps with Vorbis that one would not hear also at 256kbps.  The pre-echo of Vorbis seems to have been greatly improved since I tried it years ago.   Of course, maybe my hearing is also worse.

Well, the interesting thing about Bose, of course, alongside the ridiculously overpriced market segment that they cater to, is that they refuse to publish specifications, and the sound that their products produce is usually pretty distorted. My guess is that if you "only" hear >16 kHz on a Bose product, it's actually because the Bose product is defective and/or designed poorly and your hearing really craps out somewhere lower than that if you're listening to a speaker that faithfully represents the source material.

Re: AAC vs. Vorbis

Reply #5
Aoyumi's aoTuV is very well-tuned to the limits of human hearing on how frequencies can mask others. I thought that the lower coverage of higher frequencies than what I would expect from a psychoacoustic model at 320 kbps would result in less transparency, but I was wrong. Even at 288 kbps VBR the results came out clearer to my ears than Opus or AAC. Testing further, with spectrograms and listening tests, I found aoTuV typically beat the Apple AAC encoder on stereo fidelity. I know AAC encoders are generally weak on stereo separation where it counts.... but let's just say I'm using Vorbis more and more instead.

As far as speakers go, I have a hard time trusting speakers with amplifiers built-in. The amplifier itself is bound to cause distortion. Good headphones are better.

Re: AAC vs. Vorbis

Reply #6
I like Vorbis.  It is mature and well-balanced.  Linux support is simpler and better (especially with aoTuV) than for AAC, sound quality seems at least comparable.  Doesn't perform as well as something like Opus at very low bitrates, but effectively transparent small enough not to be a problem for me.  Device support is patchy, though, and sometimes there but not advertised.  I can't hear above 14 kHz any more, probably masks some artefacts that I could spot many years ago (unless its just better codecs).

 
SimplePortal 1.0.0 RC1 © 2008-2019