[dablues]

Hi all!
 
Been a while since I posted (new kid, new job, etc.), but missed hanging around with everyone here.
 
I was using the ABX "improved" Tidal test at http://abx.digitalfeed.net/ and got a decent score (p = 0.04 for the medium-length test), but then redid the test a couple of times and got a much worse score. However, every time I took the test (4 times), I got 100% correct for the Eagles "Hotel California" clip, and 80-100% correct for the Killers' track. I did TERRIBLY with the Daft Punk track and the James Blunt track, even though I like their music just fine (I'm not a genre snob). Dixie Chicks, I did okay.
 
Both are rock tracks, with "natural" instrumentation, and I used to be a professional rock/soul musician. I'm wondering if anyone else here has experienced heightened awareness of audio quality in genres with which they're familiar, or if they notice a heightened awareness of audio quality when it comes to how tracks are recorded (analog v digital, the decade in which it was recorded, etc.).
 
Basically, I'm trying to justify my lossless obsession. My wife can even hear the difference between Tidal and Apple Music, and she's not the audio nut that I am (and I definitely feel there's a difference).
 
Yes, I know someone will probably snarkily say "well, if you like Tidal, then just use it and stop justifying yourself." I know that argument, so don't bother! I just want to be a nerd and discuss things endlessly with my online head-fi friends. 

 

[RRod]

  Both are rock tracks, with "natural" instrumentation, and I used to be a professional rock/soul musician. I'm wondering if anyone else here has experienced heightened awareness of audio quality in genres with which they're familiar, or if they notice a heightened awareness of audio quality when it comes to how tracks are recorded (analog v digital, the decade in which it was recorded, etc.).
 

 
The only time I get positives at any decently high bitrate are killer mp3 samples, especially this one electronica track (which is certainly not my genre of familiarity). Certain tracks just hit the right buttons to give a specific codec a hard time. Getting an easy 100% at 320kbps *AAC* on common material sounds awfully suspect, so instantly I wonder what is being mucked up in this fixed Tidal test.

 

[dablues]

It's actually not the Tidal test itself, but a reworking of it, comparing 320kbps with "lossless". The sound levels and EQ are all the same, and it's an ABX. I think it's pretty legit, but maybe not. As a scientist, I trust it at this point - the methodology seems good.

 

[RRod]

  It's actually not the Tidal test itself, but a reworking of it, comparing 320kbps with "lossless". The sound levels and EQ are all the same, and it's an ABX. I think it's pretty legit, but maybe not. As a scientist, I trust it at this point - the methodology seems good.

 
Sent the samples from the two Hotel California tracks to files, and found a 0.2dB difference in RMS for the first ~12s. That's a way bigger difference than I'd expect from 320k AAC (encoding to 320k mp3 yields a 0.01dB RMS difference in the right channel for sample B, but that's about it). Anyone else care to confirm?

 

[dablues]

Okay, so maybe the methodology wasn't as good as I thought. Sorry about that. But the point of this post is to discuss whether or not people are more sensitive to audio quality based on genre, or not - the test thing was just an intro device to the larger question at hand. I appreciate your dedication to good testing - admirable! - but let's not get into validity of my results or whatever - just wanted a bigger discussion about genres and audio quality.

 

[RRod]

This thread has a good discussion of what tends to be hard to encode. In the sense that things are genre related, it's about whether or not those genres contain these types of signals. EDM will have lots of fast little bleeps that will tend to cause pre-echo problems, Classical contains solo harpsichord that has tons of harmonics that aren't easily masked, etc. I'd bet that any seasoned rock fan would have an easier time ABXing a castanet track than any rock track they've literally spent decades listening to, if we're talking higher bit rates.

 

[dablues]

Awesome! That's the info I wanted to hear. Also wanted to hear people's own experiences with quirks in genre and audio quality. THanks!

 

[sonitus mirus]

  This thread has a good discussion of what tends to be hard to encode. In the sense that things are genre related, it's about whether or not those genres contain these types of signals. EDM will have lots of fast little bleeps that will tend to cause pre-echo problems, Classical contains solo harpsichord that has tons of harmonics that aren't easily masked, etc. I'd bet that any seasoned rock fan would have an easier time ABXing a castanet track than any rock track they've literally spent decades listening to, if we're talking higher bit rates.

 
I believe that a recent release of Lame, version 3995o, was designed in an effort to help reduce the harpsichord issues.  I have not tested it yet.
 
It is available here, if anyone is interested in testing it out: Lame 3995o.

 

[RRod]

   
I believe that a recent release of Lame, version 3995o, was designed in an effort to help reduce the harpsichord issues.  I have not tested it yet.
 
It is available here, if anyone is interested in testing it out: Lame 3995o.

 
I haven't tested much lame since I went over to Opus, but all these guys keep on trucking to bring us transparency as low as they can go. Settings matter as well. In the eig.wav sample, I can ABX things with default settings at CBR 320, but not at V0 (which has slightly different options for the perceptual coding).