How To: Process High Resolution Music Files

[Music Alchemist]

Alright then, since you are willing to prove it, go here and make your case by posting about it:
 
http://www.head-fi.org/t/738552/testing-the-claim-i-can-hear-differences-between-lossless-formats
 
You will need to conduct a series of tests. We will walk you through it there.
 
But for the record...I am saying that it is physically impossible for any human to tell the difference between Red Book and high res. The only difference is different masters being used. If the same master is used, it will sound the same whether it's CD, SACD, DVD, DSD, 16-bit, 24-bit, lossless, or even 256 kbps AAC. When you convert the files yourself, from high res to Red Book, there is no audible difference, and this has been documented. You can measure it and demonstrate that any differences are below the threshold of audibility, meaning that human ears cannot hear it.
 
And no, when you compress a file from higher to lower resolution, the data is lost forever. Converting it back to higher resolution does absolutely nothing. This does not work like lossless compression does.

 

[MindsMirror]

 
Then again, your' hygene might be so bad, you have potatos growing in your' ears. Which reminds me? Seriously, when was the last time you cleaned your' ears?

In essence what you are saying is "I disagree with you on a technical matter, therefor you might have potatoes growing in your ears." That makes a lot of sense. I'll go get that checked right away.

 

[Music Alchemist]

Well lookie here! This thread has been transferred to Sound Science, where it belongs. Now we should see some real action...

 

[HiResSndWizzard]

What condition the file was before you started the upsample has everything to do with what differences will occure. If the 20k to 22.05k top of the file is erased of it's alias artifacts, I would totally agree with you.
 
I must be an impossible human, because in my 70's I can still hear 25 khz from a sine wave generator emitting an output level of -96 dB normalization into an Accuton Diamond Tweeter! I bet you can't lay claim to that.
 
I have one up on you, though. I am a communications opperator, for 60 years, & can hear extremely faint high frequencies, purely because I am aware they are present. If I were not expecting it, I wouldn't hear it any more than you. But even though you don't realize hearing it, under varying conditions, that sound will affect what you do hear in various ways.
 
I assume you are talking Red Book as in top of the line recordings, not that different other than years, refered to as RCA Red Label? I would expect there would be no difference between those two formats, IF, & only IF the alias data is wiped from the file created when the file was compressed to 44.1, & the normalization volume of the file is well below -6 dB.
 
Again you are beating a dead horse. You can't change the physics of digital sound, nor can you predict what someone else can or can not hear.
 
Another point. I can take the same master, forget another copy, or a different master of the same file, and change what it sounds like so easily, it would make your' head spin. I can turn up & down instruments, singers voices, right on down to changing the 24 dB octive roll off of the music file to some different slope, even crank the violins through the 0 dB roof if I want. If that doesn't change what the master sounds like, nothing will.
 
20 hz in the eq of the music file actually exist at -24 dB from peak 0 dB. 10 khz is nominally at -48 dB, while 20 khz is nominal at -72 dB with a noise floor as low as -132 to -144 dB.

 

[HiResSndWizzard]

wow I love that one.
 
You're quite right. It is beating a dead horse.
 
Please go to a good doc for the extraction. Your' ears are worth more than your' wallet.

 

[HiResSndWizzard]

cool, bring it on.......

 

[Music Alchemist]

I am growing weary of all this irrelevant discussion, so I am going to simplify things for you.
 
You need to do an ABX test to document whether you can distinguish between the two files. You need to do 20 trials and get at least 15 of them right for your results to be any better than random guessing. Here is a guide on how to conduct the ABX test:
 
http://www.head-fi.org/t/655879/setting-up-an-abx-test-simple-guide-to-ripping-tagging-transcoding#post_9268096
 
This test will be between a 16-bit file and another file upsampled from the original file to whichever specs you like.
 
You could also do another test comparing a 24-bit downloaded file and a 16-bit file that was converted from that 24-bit file.
 
Take the two files and conduct and publish the ABX test results here. You will need to post screenshots.
 
Also, Red Book is 16-bit / 44.1 kHz.

 

[upstateguy]

  I am growing weary of all this irrelevant discussion, so I am going to simplify things for you.
 
You need to do an ABX test to document whether you can distinguish between the two files. You need to do 20 trials and get at least 15 of them right for your results to be any better than random guessing. Here is a guide on how to conduct the ABX test:
 
http://www.head-fi.org/t/655879/setting-up-an-abx-test-simple-guide-to-ripping-tagging-transcoding#post_9268096
 
This test will be between a 16-bit file and another file upsampled from the original file to whichever specs you like.
 
You could also do another test comparing a 24-bit downloaded file and a 16-bit file that was converted from that 24-bit file.
 
Take the two files and conduct and publish the ABX test results here. You will need to post screenshots.
 
Also, Red Book is 16-bit / 44.1 kHz.

 
Come on, who wants to do all that?  How about he uploads a before and after file to one of the upload sites and we listen to it ourselves?

 

[Music Alchemist]

  Come on, who wants to do all that?  How about he uploads a before and after file to one of the upload sites and we listen to it ourselves?

 
This is about him backing up his claims. I have already done all the tests discussed here and never even perceived a difference.

 

[upstateguy]

 

  Come on, who wants to do all that?  How about he uploads a before and after file to one of the upload sites and we listen to it ourselves?

 
This is about him backing up his claims. I have already done all the tests discussed here and never even perceived a difference.

 
I can't hear a difference between a high rez file and flac from a CD, but I'd like to play around with his upsampling technique, if I could figure out how to do it.

 

[Music Alchemist]

  I can't hear a difference between a high rez file and flac from a CD, but I'd like to play around with his upsampling technique, if I could figure out how to do it.

 
Just use a program like dBpoweramp or foobar2000 and convert a 16-bit / 44.1 kHz lossless file to a 24-bit / whatever kHz lossless file. With dBpoweramp, you'll need to use uncompressed lossless to do it, though.
 
As for hi-res vs CD, that depends on the master. Some hi-res downloads sound very different from their CD counterparts; others sound the same. This is why you need to convert the files yourself to make sure you are comparing the formats and not merely different masters.

 

[titaniumgrade5]

I actually don't even resample into 32bit when EDITING audio. If all I had was an 16/44.1 file, I keep it in the 16/44.1 that it came in and don't even use dither, if all I'm doing is cutting or mixing or joining. Only if you do any effect, of mixing large portions, does 24 bit come in useful, for less noise, in EDITING. And export in such cases without dither.
 
The whole reason Audacity for example however by default resamples to 32bit while editing, and allows using a greater sampling rate, is because it doesn't matter...
 
Same when we talk about Reaper or other DAWs. And these are the sources of the music that is ultimately listened to, recordings being mixed in the DAWs.
 
Just wanted to point that in context of the whole debate. Quite often, the HD files sound better because the source had been remastered for the release. Indeed that's why I often get HD files, because they master them differently in many cases.

 

[HiResSndWizzard]

There is no way this testing could be a double blind test of the file integrity. Without my knowing the artifact data integrity has not been altered, I could not predict any outcome of upsample. I have to assume first of all this is a trick to see what I know, vs what you can hide from my knowing in this trial testing stunt.
 
You want to do it on my terms? I will record a test file at 48k24b, compress it to 44.1 16 bit, then return that file plus or minus a noise figure increase, which I can remove as well, with a special FFT I personally created. I have done it time and time again, so I have total confidence I can do it again this time.
 
The meer fact that a lot of current files available have been altered in the various areas I previously mentioned. Those files will, for what it is worth, do absolutely nothing for the upsample, leaving me looking like a fool.
 
No. You can save your' joy ride through technology lane for someone else. I have no reason to put forth the effort for no reward.
 
I suspect the possiblity, you know something is there, but don't know the real secret how upsampling is done correctly. It stares you right in the face everytime you play a file. I'm not going to be the one to show you simply for one reason. Intelectual property rights, vs you, the website, & the general public would grab it & run with it, leaving me sitting here scratching my head as to what I did wrong?
 
No, I don't buy into that deal. But, in all fairness, I will take a look at this site to see what hooks are hiding there to snag intelectual property from me.
 
I really am not into ripping tagging, but ocassionally transcode a video file.
 
Somehow I invision you sitting there wondering just what I might know, that you don't. Considering the hidden secret I found years ago, relating to upsampling from a compression to a higher res file, I am not so sure anyone else has ever thought of it. It is for sure I have never told, or explained to anyone over the years what I found by pure accident.
 
Besides that fact, there are other artifacts that come with the upsample, which is in some ways just as bad as other unwanted artifacts. I wrote an FFT for the removal of those artifacts.
 
When I get done with the file, it will sometimes sound better than the original master, considering some of the same artifacts are in the original master, along with the new ones created during the upsample.
 
And, yes, you can call upsampling the same thing as transcoding to some extent, but that is where the simularity ends. There is a lot more to it than just resampling with pro tools. That is as far as I am willing to go to explain the technic.

 

[HiResSndWizzard]

I will consider that option, but first I have work, then I will play around with that idea.

 

[Music Alchemist]

  There is no way this testing could be a double blind test of the file integrity. Without my knowing the artifact data integrity has not been altered, I could not predict any outcome of upsample. I have to assume first of all this is a trick to see what I know, vs what you can hide from my knowing in this trial testing stunt.
 
You want to do it on my terms? I will record a test file at 48k24b, compress it to 44.1 16 bit, then return that file plus or minus a noise figure increase, which I can remove as well, with a special FFT I personally created. I have done it time and time again, so I have total confidence I can do it again this time.

 
If you cannot pass an ABX test of the files that have been properly converted (and not edited in any way) to the various resolutions, then we're done here.