Dont worry about it, it just proves we all hear different things. If you happy with the Melco thats all that matters and its fulfilled its remit. Thanks for allowing us to experiment and discover things in the process and a merry Christmas to you as well.
Seeing as myself and David are lucky enough to be from the best country in the world 
, I should be able to add to this debate in the new year when David visits with a USB stick.
Hopefully by then I might have some feedback from Dr. Aixâs high res test.
Iâm not expecting that I can conclusively hear a difference between high res and cd quality but some of the choices are mp3 apparently so Iâll have an idea whether my musical memory is actually up to discerning differences if I can or canât accurately tell the difference.
Iâd recommend as many of us as possible take this test. We seem to want to benchmark everything except ourselves.
.sjb
Just signed up.
CristalGipsy PMâed me his guesses (and he really did emphasize that they were guesses). He got 0/4 correct.
Along with Davidâs 1/4 correct, the hypothesis that there is a clear-as-a-bell audible difference between the two rips is looking a bit tattered. But, since we already established that the two rips were bit-for-bit identical, that hypothesis was pretty silly to begin with.
We could try again, with @Rob_BrownGHFâs rips, which he claims are audibly different. But he never responded to my PM, so I guess heâs rethought those claims.
1 Like
My apologies for not responding to your private message. I hope you understand that I am not glued to the forums. I will try to find two distinct rips from two different rippers but this will take time. I appreciate youâre eager to prove your point, but donât shame me within 24 hours for not replying. The forums are here to share a mutual enjoyment of music and Roon, not prove to everyone that youâre right and everyone else is wrong, @Jacques_Distler.
My profuse apologies. Take as much time as you want. Or not, as the case may be. You did express a reluctance to âchip inâ to this discussion. So youâre certainly under no obligation to participate âŚ
Same CD, different rippers. Do you still want me to email the files to you?
Equipment used - Naim Uniti Core and D100 into a Mac through dBpoweramp.
note, the date appears very unusual, but I assure you the LOANUNITICORE version was ripped moments before the screenshot was taken.
Yes, if you donât mind.
Thanks!
Letâs go through the same steps as before. First, as Rob notes, the files obviously have radically-different sizes:
% ls -l '06 - Money.flac' '06 Pink Floyd - Money.flac'
-rw-r--r-- 1 distler 49870404 Apr 6 2019 '06 - Money.flac'
-rw-r--r-- 1 distler 67646879 Jun 15 2019 '06 Pink Floyd - Money.flac'
This is for two reasons
- FLAC is a (lossless) compressed audio format. If you (or, more precisely, your ripper) use different compression settings, the resulting files will have different sizes. We canât really compare them until we uncompress the data.
- The metadata is, of course, different. So we need to remove the metadata before we can compare the files.
% ffmpeg -y -i '06 - Money.flac' -map_metadata -1 a.wav
% ffmpeg -y -i '06 Pink Floyd - Money.flac' -map_metadata -1 b.wav
Now, theyâre exactly the same size:
% ls -l a.wav b.wav
-rw-r--r-- 1 distler 67516590 Dec 20 10:15 a.wav
-rw-r--r-- 1 distler 67516590 Dec 20 10:16 b.wav
At this point, with Davidâs files, we had a problem: there was a 12ms offset in the way they were padded with silence. In order to compare them, we had to strip the leading/trailing silence. With Robâs files, weâre luckier. The time-alignment is the same, so we can just skip that step.
% sha256sum a.wav b.wav
064afa4d6a058dc5f20702a44ebd94e5e6f903c8d9b43e6e50c4b09c6ab561da a.wav
064afa4d6a058dc5f20702a44ebd94e5e6f903c8d9b43e6e50c4b09c6ab561da b.wav
As expected, they are bit-for-bit identical. Unless one of them was a bad rip, this was a foregone conclusion. They could not have been anything but identical.
Still, sha256 hashes might not convince the sceptic. Perhaps Rob works for the NSA and has found a way to break sha256.
So letâs look at the files in Audacity. Here are the two tracks:
Now we invert the second track:
Finally, we choose Tracks â Mix â Mix and Render:
Total silence. Either Rob doesnât work for the NSA, or they havenât broken sha256. But, either way, yet another confirmation that the two rips are identical.
In choosing between the two rippers, the only difference was the size of the FLAC file produced. Smaller is better, so I suppose the D100 wins. But, given the price of disk space these days, I wouldnât pay more than $100 for better FLAC compression.
4 Likes
In these original playback situations where differences are noted, I assume that the ReplayGain tag values (if RG tags are used) are identical between the compared rips and/or the volume leveling data added by Roon is identical for both files.** As everyone here knows Iâm sure, even tiny differences in playback volume can affect a userâs preference for one file over the other.
**EDIT: Or Roon setting to use RG values or volume leveling is turned OFF.
1 Like
Your theory is sound. However you did not listen to the two recordings, they are unquestionably sonically different. Confirmed on a Hugo 2 and Sennheiser HD820. Of course if you pre render and unpack the two files then they will perform identically. However this is not how streamers work, they render a few seconds in advance.
Who else can hear the difference? https://www.dropbox.com/sh/b84ilck2lee00ws/AABP8D9-QSGiqSEg3NdjmhMIa?dl=0
Thatâs how RAAT works. It sends uncompressed PCM down the wire.
Of course, even the lowly Raspberry Pi3Bâs that I use as streamers have more than enough horsepower to uncompress FLAC in real-time, without working up even the slightest bit of a sweat.
As I did for David, Iâll be happy to send you links to the (uncompressed, metadata-removed) files, and you can see whether you can distinguish them. But I think weâre agreed that would be silly.
FWIW, in response to @garym, neither of Robâs files have ReplayGain tags, so thatâs not why he hears a difference between them.
Even without RG tags, canât Roon add its own volume leveling information it creates from analyzing the files upon import and places in the Roon database (and use this information in playblack)?
edit: just trying to find some plausible reason as to why two identical files could sound VERY different to these folks.
1 Like
Oh, sure. But since the PCM data is identical, the volume-leveling info will be the same.
2 Likes
I appreciate the effort youâve gone to. Strip Roon and all processing out of the equation and listen. Donât analyse, but listen to it. As you have correctly demonstrated, apply enough processing you will unravel the same zeroes and ones in the uncompressed file. That is not the same principle as âthey audibly sound differentâ.
With all due respect, I am not contributing any further to this topic. Iâve said my piece, youâve provided your assessment and we can walk away from this with our dignity somewhat intact.
Two identical files CANNOT sound different other than the listenerâs preconceived bias.
We live in the golden age of processing power (and, hence, in the golden age of digital audio). A $35 Raspberry Pi3B has (vastly) more processing power than the computers that sent men to the moon.
Nowadays, whatâs hard is avoiding the temptation to throw too much processing power at a task. I deliberately run Roon Core on a machine that is under-specced, relative to Roonâs recommendations. Roon Core runs great, so I figure that thereâs no point in spending more money and wasting more energy (TDP = 10W, idles at 4W) for no discernible benefit.
Youâre missing the point. The more it cost, the better it sounds. How could it not?
Exactly what Iâve been wondering.
Where can I find the tag info for ReplayGain in a trackâs metadata?
Enjoying this thread and learning a bit.â
Iâm a Mac person, so I use Metadatics to edit metadata tags. There are plenty of Windows alternatives, but I wouldnât know what to recommend.
1 Like