24bit vs 16bit, the myth exploded!

[kraken2109]

Lol I can abx mp3 from flac with accuracy.

Can you prove it? A lot of people make this claim but most can't.

 

[OddE]

Can you prove it? A lot of people make this claim but most can't.

-He didn't specify the bit rate of the mp3.

Heck, even I can tell the difference between flac and a 64kbps mp3 of the same recording...

 

[Soundsgoodtome]

320kbps ogg Vorbis (better than mp3 coding I'm told) vs cd flac. People make the claim sure but I can honestly say I heard it on my HE560 rig, the latter is better and it's more noticeable in the treble.

How does one prove it with definitive proof over a forum? Just because people say it without proof doesn't automatically mean they're lying, you are entitled to your opinion of course and ymmv yada yada.

Can you prove it? A lot of people make this claim but most can't.

-He didn't specify the bit rate of the mp3.

Heck, even I can tell the difference between flac and a 64kbps mp3 of the same recording...

 

[Head Injury]

320kbps ogg Vorbis (better than mp3 coding I'm told) vs cd flac. People make the claim sure but I can honestly say I heard it on my HE560 rig, the latter is better and it's more noticeable in the treble.

How does one prove it with definitive proof over a forum? Just because people say it without proof doesn't automatically mean their lying, you are entitled to your opinion of course and ymmv yada yada.
 

Download foobar2000 if you haven't yet, find and download the ABX Comparator plugin, do 10 or ideally 20 trials comparing the OGG to FLAC, and post your results. For lack of a more controlled test, I'll take your word for it if you pass this with confidence.

 

[Soundsgoodtome]

If I don't get it right I'll post it as well, never used abx but to compare Redbook flac vs hi -res. I have compared 320 Spotify offline to cd rips. I've got all the software but do need to convert some FLAC files to ogg Vorbis. Probably do the abx tomorrow in the a.m.

 

[U2Bono269]

What percentage do you all consider to be passing, or sufficient? Do you have to get it right 100% of the time? Or just a majority?

 

[RRod]

  What percentage do you all consider to be passing, or sufficient? Do you have to get it right 100% of the time? Or just a majority?

 
It all depends on what you want your false positive rate to be. If you are fine with 5% of tests* yielding a false positive, then 9/10 is considered enough. Here are the necessary correct trials (for 5% false-positive rate) shamelessly stolen from Wikipedia:

Number of trials​	10​	11​	12​	13​	14​	15​	16​	17​	18​	19​	20​	21​	22​	23​	24​	25​
Minimum number correct	9	9	10	10	11	12	12	13	13	14	15	15	16	16	17	18

 
*Note that a "test" means the entire set of trials you have decided to do beforehand

 

[Soundsgoodtome]

I would think a 66-75% accuracy would be considered sufficient.

 

[U2Bono269]

   
It all depends on what you want your false positive rate to be. If you are fine with 5% of tests* yielding a false positive, then 9/10 is considered enough. Here are the necessary correct trials (for 5% false-positive rate) shamelessly stolen from Wikipedia:

Number of trials​	10​	11​	12​	13​	14​	15​	16​	17​	18​	19​	20​	21​	22​	23​	24​	25​
Minimum number correct	9	9	10	10	11	12	12	13	13	14	15	15	16	16	17	18

 
*Note that a "test" means the entire set of trials you have decided to do beforehand

I didn't particularly care about this before, but I think I'm going to do this again with more variety. I did it once before, and I ran 2 trials of 10. I scored a 9 and an 8, but I didn't save the logs. I assumed that meant I passed. This was between a FLAC and a 256 AAC.
 
I'll do some tests with 16 and 24 bit too, because why not. I've got free time this week.

 

[RRod]

 
I didn't particularly care about this before, but I think I'm going to do this again with more variety. I did it once before, and I ran 2 trials of 10. I scored a 9 and an 8, but I didn't save the logs. I assumed that meant I passed. This was between a FLAC and a 256 AAC.
 
I'll do some tests with 16 and 24 bit too, because why not. I've got free time this week.

 
The 9 is a pass; the 8 isn't (at the 5% level), but there's a lot more to doing a correct ABX test (level matching, etc) that can be automated by software. What are you going to do these tests in?

 

[U2Bono269]

   
The 9 is a pass; the 8 isn't (at the 5% level), but there's a lot more to doing a correct ABX test (level matching, etc) that can be automated by software. What are you going to do these tests in?

Foobar with the ABX comparator plugin. Doesn't that automatically level match?

 

[RRod]

 
Foobar with the ABX comparator plugin. Doesn't that automatically level match?

 
Yep that should do.

 

[U2Bono269]

That's what I used last time. Very easy.

 

[Soundsgoodtome]

So now that we're derailed I'll ask a question for further derailment. The 9/10 accuracy test and such but wouldn't the music playing also play a rile? I'd presume taking this test would mean using music with quality recordings and perhaps nothing too simple musically.

 

[bigshot]

Complexity of the music isn't the issue. There are just certain kinds of sounds that are more prone to artifact. Transparency is the lack of momentary artifacts. It isn't an overall sound quality. It's that the higher the rate, the fewer artifacts there are, until the artifacts disappear entirely.