[monterto]

I'm a scrub when it comes to HI-FI and the like, but I'm doing my best to get my feet wet. I run a set of shure SRH 840s and listen to my music through my laptop (plan on buying a DAC soon) and I have noticed an audible difference between my 320 and FLAC music. Now what confuses me is that these frequencies which are beyond my hearing make a difference in the sound quality of the song. In my mind it makes sense that 320 CBR would have equal sound to FLAC, but this just ins't the case. Can any of you gurus explain this to me?
 
Cheers!
 
TL;DR Why can I hear a difference between FLAC and 320

 

[jasonb]

Sounds like a bad or old mp3 encoder. 320kbps should sound just fine. 320kbps is what most of my listening is done with.

 

[monterto]

I encode my own 320s using the latest version of LAME, Maybe I'm imagining the difference...

 

[eucariote]

Were you doing your comparisons sighted?  If so I'd chalk it up to expectation (unless your encoders record at different volumes, etc.).  Try it again blind, for 20 trials and post your results.  I'll bet your results aren't significantly different from random.

 

[violinvirtuoso]

But hey, if you use the ABX Comparator and can consistently can tell the difference, you can deem yourself as having golden ears (or close enough).

 

[monterto]

I did a very small and brief blind test with my sister where I encoded a song to V2, V0 and 320 and had her play me all the MP3s and the FLAC in a random order and I did a thumbs up or down comparing the SQ of each. I thought the MP3s sounded more sibilant in the highs than the FLAC but that might have just been a guess.
 
Maybe once I have a proper DAC, I'll do some more blind tests and post my  results.

 

[Ham Sandwich]

Make sure both the MP3 and the FLAC files are getting played the same way.  Computers can do strange and unexpected things like play MP3 through a different audio path than FLAC depending on what drivers and codecs you have.
 
Use Foobar2000 as a sanity check.  We know it does the right thing and will the audio through the path you have specified in the playback options.  And we know it plays MP3 and FLAC correctly with no funny business.

 

[monterto]

Foobar is my go-to media player on my computer so that shouldn't be a problem. I also just downloaded the blind test plug-in that eucariote reccomended.

 

[Ham Sandwich]

What music are you testing with?  Is it music that has been loudnessed?  Songs that have gotten the loudness treatment can flirt with 0 dBFS and can attempt to achieve the impossible of wanting to go beyond 0 dBFS.  Here's a little bit of info about 0 dBFS: Issues with 0dBFS+ Levels On Digital Audio Playback Systems.  Some DACs don't handle the 0 dBFS+ levels well while some DACs do better.
 
A hypothesis could be that MP3 decoding is handling 0 dBFS levels (or near 0 dBFS levels) in a way that is better for your DAC.  When MP3 is decoded the 0 dBFS+ and near 0 dBFS levels can get treated differently than pure lossless (FLAC).  MP3 decoding can actually exceed 0 dBFS while FLAC decoding cannot.  And it would depend on how the software playback processing treats those situations and how your DAC treats those situations.

 

[monterto]

 
Quote:

What music are you testing with?  Is it music that has been loudnessed?  Songs that have gotten the loudness treatment can flirt with 0 dBFS and can attempt to achieve the impossible of wanting to go beyond 0 dBFS.  Here's a little bit of info about 0 dBFS: Issues with 0dBFS+ Levels On Digital Audio Playback Systems.  Some DACs don't handle the 0 dBFS+ levels well while some DACs do better.
 
A hypothesis could be that MP3 decoding is handling 0 dBFS levels (or near 0 dBFS levels) in a way that is better for your DAC.  When MP3 is decoded the 0 dBFS+ and near 0 dBFS levels can get treated differently than pure lossless (FLAC).  MP3 decoding can actually exceed 0 dBFS while FLAC decoding cannot.  And it would depend on how the software playback processing treats those situations and how your DAC treats those situations.

This is FANTASTIC info. THANKS!
 
I will have to do a bit more reading to fully understand it though.

 

[eucariote]

You can get a quick and dirty statistical probability score of your results in this website.
 
For the 'sample size', enter the number of trials that you took in your test (do an even number for simplicity).  For 'expected', enter 1/2 of that number, which is the null hypothesis that your guesses will be purely random and thus 50% correct (out of two possible outcomes) as to which sample rate your were hearing.  For 'observed', enter the actual number of correct guesses that you made.  You will get a probability score of the likelihood of your correct responses given the null hypothesis that your score will be purely random.  A probability score of <.05 means that your score was very unlikely (so not random) given the null hypothesis, so the null hypothesis can be rejected and you can accept the alternate hypothesis that you can/did hear a difference.

 

[dex85]

a little advice, don't do too many trials a day. i achieved much better results by doing just two trials a day and calculating probabalities later. i tend to get confused when i take more trials or when i do rapid switching between samples. i usually cut samples to 30 seconds. it has worked best for me, but we are all differenent, so you may try other approach to get best results.