[Stereodude]

Quote:

Originally Posted by Head Injury /img/forum/go_quote.gif
 
You can take a 128 kbps MP3 and convert it to a 320 kbps AAC, and it'll sound worse.

So you have the ABX results to prove this claim?

 

[Head Injury]

Quote:

So you have the ABX results to prove this claim?

No, because that's how lossy works. There's no guarantee anyone can tell the difference (so perhaps I shouldn't have used the word "sound"), but there will be missing information and it will be objectively worse.
 
A lossy codec automatically discards portions of the audio it's fed according to how it's coded. It doesn't know if it's being fed a lossy file already. It sees data, it removes and alters it.

 

[nick_charles]

 
Quote:

No, because that's how lossy works. There's no guarantee anyone can tell the difference (so perhaps I shouldn't have used the word "sound"), but there will be missing information and it will be objectively worse.
 
A lossy codec automatically discards portions of the audio it's fed according to how it's coded. It doesn't know if it's being fed a lossy file already. It sees data, it removes and alters it.

It isnt automatic, it is based on a set of psychoacoustic algorithms, it only throws away data that its model says cannot be perceived, but if that data such as proximal frequencies that would normally be masked has already been discarded the algorithms may not have anything left to delete.
 
This is a testable hypothesis. What one could do is take a wav file, encode to mp3, convert back to wav using Lame then repeat the encoding, rinse and repeat then do a spectral analysis on original, 1st gen, 2nd gen and so on....could be an interesting exercise

 
 

 

[khaos974]

Quote:

Quote:

No, because that's how lossy works. There's no guarantee anyone can tell the difference (so perhaps I shouldn't have used the word "sound"), but there will be missing information and it will be objectively worse.
 
A lossy codec automatically discards portions of the audio it's fed according to how it's coded. It doesn't know if it's being fed a lossy file already. It sees data, it removes and alters it.

It isnt automatic, it is based on a set of psychoacoustic algorithms, it only throws away data that its model says cannot be perceived, but if that data such as proximal frequencies that would normally be masked has already been discarded the algorithms may not have anything left to delete.
 
This is a testable hypothesis. What one could do is take a wav file, encode to mp3, convert back to wav using Lame then repeat the encoding, rinse and repeat then do a spectral analysis on original, 1st gen, 2nd gen and so on....could be an interesting exercise

Just tried, 10 iterations of Lame V5 (~130 kbps), the result is utterly horrible, 1st gen is ok in a standard mp3 way, 2nd gen sounds already quite crappy with artifact popping up here and there, 10th gen is horrible, absolutely horrible.
 
Here's a shot between 1st and 10th gen, you can see the added high frequency distortion easily enough.
 

 
 

 

[RexAeterna]

Hahaha, yes. Totally agree with that statement. 

i don't. there is some nice sounding stuff on youtube. you guys over exaggerate too much.

 

[Stereodude]

Quote:

Just tried, 10 iterations of Lame V5 (~130 kbps), the result is utterly horrible, 1st gen is ok in a standard mp3 way, 2nd gen sounds already quite crappy with artifact popping up here and there, 10th gen is horrible, absolutely horrible.
 
Here's a shot between 1st and 10th gen, you can see the added high frequency distortion easily enough.

But you didn't try turning a 128kbit mp3 into a 320kbit AAC file and demonstrate it was audibly different.

 

[khaos974]

Quote:

Quote:

Just tried, 10 iterations of Lame V5 (~130 kbps), the result is utterly horrible, 1st gen is ok in a standard mp3 way, 2nd gen sounds already quite crappy with artifact popping up here and there, 10th gen is horrible, absolutely horrible.
 
Here's a shot between 1st and 10th gen, you can see the added high frequency distortion easily enough.

But you didn't try turning a 128kbit mp3 into a 320kbit AAC file and demonstrate it was audibly different.

It wasn't my intention, in fact, I think that a single conversion from 128 kbps mp3 to 320 aac would be very difficult to detect in ABX.
 
 

 

[nick_charles]

Quote:

Just tried, 10 iterations of Lame V5 (~130 kbps), the result is utterly horrible, 1st gen is ok in a standard mp3 way, 2nd gen sounds already quite crappy with artifact popping up here and there, 10th gen is horrible, absolutely horrible.
 
Here's a shot between 1st and 10th gen, you can see the added high frequency distortion easily enough.
 

 
 

Thanks for doing that, interesting results.
 

 

[Austin Morrow]

Quote:

No, because that's how lossy works. There's no guarantee anyone can tell the difference (so perhaps I shouldn't have used the word "sound"), but there will be missing information and it will be objectively worse.
 
A lossy codec automatically discards portions of the audio it's fed according to how it's coded. It doesn't know if it's being fed a lossy file already. It sees data, it removes and alters it.

Exactly, Head Injury. 128 Kbps, even if converted to FLAC, will still be lossy, it's just the way it is. Now, two different things going on here however. If you were to convert a 128 Kbps to FLAC, then no data would be lost since FLAC is lossless, it;s still stay at 128 Kbps. If you were to convert 128 Kbps to MP3, AAC, or any other of the music files, then you would be losing music data every time you compress and convert.

 

[Stereodude]

Quote:

If you were to convert 128 Kbps to MP3, AAC, or any other of the music files, then you would be losing music data every time you compress and convert.

But, if it the loss is not audible, does it really matter?

 

[Head Injury]

Quote:

But, if it the loss is not audible, does it really matter?

Well, you don't want to damage your digital files, do you? I can't tell between high bitrate MP3 and lossless, but I rip to lossless whenever possible. Just look at the artifacts khaos974 created.
 
You can try for yourself, if you like. I'm not concerned enough.

 

[Austin Morrow]

Quote:

Well, you don't want to damage your digital files, do you? I can't tell between high bitrate MP3 and lossless, but I rip to lossless whenever possible. Just look at the artifacts khaos974 created.
 
You can try for yourself, if you like. I'm not concerned enough.

Same here. Can't hear a difference between 320 Kbps and Lossless unless on an HD800. I have OCD and I just feel better when I rip in lossless.

 

[Stereodude]

Quote:

Well, you don't want to damage your digital files, do you? I can't tell between high bitrate MP3 and lossless, but I rip to lossless whenever possible. Just look at the artifacts khaos974 created.
 
You can try for yourself, if you like. I'm not concerned enough.

I rip all my music to lossless (FLAC).  I do it mostly because I don't want to re-rip to get the files into another format in the future.  That and HD space is relatively cheap.  However for putting music on a portable music player I have no problems with lossy compression.  It's not the devil most people make it out to be.

 

[Head Injury]

Quote:

I rip all my music to lossless (FLAC).  I do it mostly because I don't want to re-rip to get the files into another format in the future.  That and HD space is relatively cheap.  However for putting music on a portable music player I have no problems with lossy compression.  It's not the devil most people make it out to be.

I know. That's exactly what I do, and I regularly challenge dissenters to blind tests. However, you still want to minimize the damage, and that means high bitrate (I use V0, V2 minimum) and no re-encoding lossy to lossy (since that saves no space vs. lossless to lossy, and does additional damage).

 

[Austin Morrow]

Quote:

I rip all my music to lossless (FLAC).  I do it mostly because I don't want to re-rip to get the files into another format in the future.  That and HD space is relatively cheap.  However for putting music on a portable music player I have no problems with lossy compression.  It's not the devil most people make it out to be.

I don't care about lossy compression filed on my iPhone at all. Almost all of my IEM's (Grado GR10 - e-Q5) aren't revealing enough to show the effects of a lossy compression of 256-320 Kbps AAC file.