Skip to main content
Topic: Am I using (generally accepted) transparent settings? (Read 364 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Am I using (generally accepted) transparent settings?

I had an discussion with a friend that made me doubt my AAC settings choice...

My music collection is ripped to FLAC and AAC (AAC is for my car and portable devices). To encode my FLAC & WAV to AAC I'm using QAAC 2.68 and this is the command line I'm using:

Code: [Select]
-v192 -q2 --tag="too:" *.wav
  • Constrained VBR mode set to 192k
  • Best (2) AAC encoding Quality
  • Strip the codec details

Are these settings producing transparent AAC files (at least considered transparent for most)? Am I missing something that could improve my encodings?

Re: Am I using (generally accepted) transparent settings?

Reply #1
I had an discussion with a friend that made me doubt my AAC settings choice...
[…]
Are these settings producing transparent AAC files (at least considered transparent for most)? Am I missing something that could improve my encodings?
For most people, 192 kbps is transparent for decades. With a mature AAC encodings people won't have any issues other than placebo or audiofoolic neurosis. Even with high quality hardware. Especially with high quality hardware…
If you want to be sure, do some ABX tests: it's what you hear that matters, not what your friend hear or think about it.

You can improve theoretical quality by increasing the bitrate: the same AAC encoder will produce a better objective output at 256 or 320 kbps than with 192. But will subjective quality increase as well? I doubt it. Again, answer is in ABX procedure.

Re: Am I using (generally accepted) transparent settings?

Reply #2
I remember seeing some years back a site in which some guy compared with charts the waveforms from lossless vs. various compressed setting, showing what information was lost when compressing. Anyone know where that might still be?

Re: Am I using (generally accepted) transparent settings?

Reply #3
I remember seeing some years back a site in which some guy compared with charts the waveforms from lossless vs. various compressed setting, showing what information was lost when compressing. Anyone know where that might still be?
Such pictures won't give you an accurate idea about what is truly lost by the encoding process. The only thing you can really see is the lowpass (i.e the lack of information on higher frequencies) which is often seen by many people as as the only missing information. But it's quite inexact because a lossy encoder does much more than simply lowpassing: it changes almost everything and the missing sounds are not always visible nor audible.
For example, low bitrate HE-AAC encoding start to resample everything at half sampling rate (44100 Hz is resampled to 22050 Hz, which means no information above 11 KHz). On decoding the original sampling rate is restored, and the missing information on higher frequencies are reconstructed from almost nothing. On a spectral or graphical picture it seems that most information are still maintained by the format but in reality they are really different information from the original. And to the ears the result is often far from transparency.

So forget those charts. They can't be used for measuring quality of any lossy encoder.

 
SimplePortal 1.0.0 RC1 © 2008-2020