Head-Fi.org › Forums › Equipment Forums › Sound Science › 320 kbps vs FLAC?
New Posts  All Forums:Forum Nav:

320 kbps vs FLAC? - Page 3

post #31 of 51
Quote:
Originally Posted by Pistachio View Post
Yeah, that's the thing, you THINK you hear a difference. I am telling you that you don't ACTUALLY hear one. I have been involved with audiophoolery for a very long time now, I have read these forums since 2004, I have seen this topic come up a million times, and I have heard from people exactly like you who have never ABX'ed 320 vs Lossless and claim to be able to tell the difference, I used to be one of these people myself.
I can't really hear the difference from my 192kbps recording of Linkin park's Papercut from my 320kbps one yet I can a slight increase in quality from my Flac compare to the 320kbps, this placebo effect you so sure we all have should made me think I can hear a noticeable difference between the 192 and 320, how can you be so sure the differences people hear from Flac compare to a 320 is placebo when scientifically Flac is better and has more audio bits? people can't pass some quick A/B testing means that any other type of listening is Placebo?
post #32 of 51
Quote:
Originally Posted by Fortgammon View Post
I can't really hear the difference from my 192kbps recording of Linkin park's Papercut from my 320kbps one yet I can a slight increase in quality from my Flac compare to the 320kbps, this placebo effect you so sure we all have should made me think I can hear a noticeable difference between the 192 and 320, how can you so sure the differences people hear from Flac compare to a 320 is placebo when scientifically Flac is better and has more audio bits? people can't pass some quick A/B testing means that any other type of listening is Placebo?
So, because you think there is no change between 192 and 320 but a small change between 320 and FLAC I should formulate a theory as to why you experience things in this way?

Why don't you just go and ABX in foobar.
post #33 of 51
Quote:
Originally Posted by Fortgammon View Post
I can't really hear the difference from my 192kbps recording of Linkin park's Papercut from my 320kbps one yet I can a slight increase in quality from my Flac compare to the 320kbps, this placebo effect you so sure we all have should made me think I can hear a noticeable difference between the 192 and 320, how can you so sure the differences people hear from Flac compare to a 320 is placebo when scientifically Flac is better and has more audio bits? people can't pass some quick A/B testing means that any other type of listening is Placebo?
Just because it's better in theory doesn't mean it's better aurally. I've been able to ABX v0/320, but try as I might, I can never hear anything even remotely different between 320 and FLAC without a "killer sample". And to your last question, seriously, the only way to be free of the placebo is ABX, or at east a good double blind test.
post #34 of 51
I ABX'd My 320kbps and Flac version of Linkin Park's Papercut, I did multiple trials

Score 5/7
Guess Prob: 22.7%

I admit it wasn't easy, though the trails I failed at I rushed, the ones I got correct took me more time going back and forth through the samples. Papercut in Flac has never in my mind been very different versus it's 320kbps version even compare to other Flacs I've downloaded,
and it's one of my most listen to songs

This ABX option is quite entertaining I'm gonna use again with the Flac and 320 Disturbed's Indestructible

Edit: Damn I got a low score for Indestructible

Score 3/8
Guess Prob: 85.5%

this is a 2008 heavy metal song so I assumed it's been "made louder" aka screwed a few times by the time it got to a CD

But Perfect score for KoRn's Right Now

Score 5/5
Guess Prob: 3.1%
post #35 of 51
"Better sound quality" - As in more transparent to the source, or closer to your preference in sound?
If the former, then FLAC take a clear win since it is lossless. But if the later then you may actually prefer the sound of 320kbps (lossy encoded) file. The phsycoacustic algorithm involved may "smoothen" or "take the edge off" the sound to your liking.

Besides sheer sound quality there are other reasons to go either lossy or lossless as well.
post #36 of 51
Quote:
Originally Posted by krmathis View Post
"Better sound quality" - As in more transparent to the source, or closer to your preference in sound?
If the former, then FLAC take a clear win since it is lossless. But if the later then you may actually prefer the sound of 320kbps (lossy encoded) file. The phsycoacustic algorithm involved may "smoothen" or "take the edge off" the sound to your liking.

Besides sheer sound quality there are other reasons to go either lossy or lossless as well.
Krmathis,

Do you have or know of any examples where people thought going lossy on a file improved the sound aesthetically? I'm not questioning or challenging it at all - I'm just really curious as to how this may have been tested and what "benefits" were experienced.
post #37 of 51
^ No links out of my head right now.
But I have seen statements around here about members who prefer lossy (think it was MP3) over lossless in regards of sound quality. Bad records may have noise, clicks, ... which I believe a lossy encoder may mask out while throwing away audio data to save space.
post #38 of 51
Quote:
Originally Posted by Pistachio View Post
Yeah, that's the thing, you THINK you hear a difference. I am telling you that you don't ACTUALLY hear one.
Well, that's a hell of an assertion.

Quote:
I have been involved with audiophoolery for a very long time now, I have read these forums since 2004,
And you've only posted sixty-two times? I gotta admire that, actually...kinda zen, almost.
Quote:
This isn't even contentious. The figure isn't even close to 50%. It isn't too far from 99% of people can't tell. The 1% that have trained themselves to hear the difference can only do so on very very specific types of music. Close to 100% of people can tell the difference on "killer samples" but these are extremely rare, with only a small group of such samples being known.
I'm a bit curious...what sort of equipment are you assuming these people are using? I mean, I definitely couldn't tell the difference on an iPod with stock buds, but if you gave me an O2 with a Blue Hawaii and a world-class DAC, I'd probably at least have a shot. I'm guessing that these figures aren't based on those extremes, but it's pretty indisputable that different equipment can be more or less forgiving of such things.

I think we can all agree, however, that FLAC is indisputably more portable/transcodable than 320, not that much bigger, and generally superior for collection/archival purposes. So, perhaps it doesn't have immediate auditory payoff, but with hard drives expanding I'd say that at least 16/48 FLAC has enough going for it that debating the auditory difference has become mostly academic.

Oh, and 192 to 320? Incredibly audible ...can't listen to my old eMusic MP3s anymore except on the go.
post #39 of 51
it may be because i have a very sensitive ear, but my 320 tracks dont have the depth my FLAC do. Idk how to explain it, it just sounds on a slightly higher plane of existence
post #40 of 51
I love flac and I find a huge difference between it and 320 mp3. I don't find any difference between 320 and v0 though :/ Anyway I don't usually hear things in flac that i don't hear lossy files but I find a huge difference in the soundstage (I guess that's the word?) and the brightness and depth of the music. If I do hear something else in flac that i don't hear in lossy encodes, it's usually because of the improved soundstage.
post #41 of 51
Quote:
Originally Posted by danroche View Post
Vel, thanks for posting that study. It was really interesting and I appreciate you kicking the real deal with how your findings varied between sample selection, bitrates, etc., not to mention referencing the P values for each test.

Do you think the results would be different were you to re-perform the test with LAME 3.98 or AAC at higher or equivalent bitrates? I know that the MP3 encoders packaged with iTunes have been left to rot a bit since AAC came on the scene back in 2003, but it's very possible that using a different encoder wouldn't provide any noticible improvements at higher bit rates.
You're welcome I tried to bring a bit of science into the whole issue and was very pleased with my results.

You are absolutely correct, different encoders might have much better encoding which would in fact be indistinguishable. My results are obviously only applicable to the precise version of the iTunes encoder I specified in the posts. The only argument I can make is that many people have music libraries that are not created with the perfect encoder at the perfect settings and the reality is that many people do just use iTunes to rip their music. So as a 'real world' test I believe it is accurate. As for someone who wants to rip their CD's once-and-for-all, they could find the perfect encoder and settings but chances are in a few years time something better will come along so why not just rip in FLAC to begin with.

I must also disagree with another post you made regarding the difference only being heard in certain samples. For me, the thing that instantly gave away a 320 vs a flac was the snares. The clarity of a flac snare was instantly recognisable and I now tend to notice it more in my lossless music. Since most music has snares or other very short sharp sounds, it is not difficult to distinguish 320 and flac for almost any sample as long as you know what to look for. I must confess that that is the primary difference I hear between the samples so I'm sure there are other differences I'm not picking up as well.
post #42 of 51
Quote:
Originally Posted by Vel View Post
You're welcome I tried to bring a bit of science into the whole issue and was very pleased with my results.

You are absolutely correct, different encoders might have much better encoding which would in fact be indistinguishable. My results are obviously only applicable to the precise version of the iTunes encoder I specified in the posts. The only argument I can make is that many people have music libraries that are not created with the perfect encoder at the perfect settings and the reality is that many people do just use iTunes to rip their music. So as a 'real world' test I believe it is accurate. As for someone who wants to rip their CD's once-and-for-all, they could find the perfect encoder and settings but chances are in a few years time something better will come along so why not just rip in FLAC to begin with.

I must also disagree with another post you made regarding the difference only being heard in certain samples. For me, the thing that instantly gave away a 320 vs a flac was the snares. The clarity of a flac snare was instantly recognisable and I now tend to notice it more in my lossless music. Since most music has snares or other very short sharp sounds, it is not difficult to distinguish 320 and flac for almost any sample as long as you know what to look for. I must confess that that is the primary difference I hear between the samples so I'm sure there are other differences I'm not picking up as well.
Vel,

I would submit a couple of things for consideration/further research here, and this is all in good fun:

1. Reading over the comment about snares, I do believe that these might have been artifacts from using the MP3 encoder in iTunes, largely because listening-tests, even among informed listeners, haven't observed these kinds of regularly occuring artifacts in v0 and 320kbps LAME MP3s (and believe me I've personally looked for them.) I do recall, and given this was several years ago, noticing iTunes-encoded MP3s having a particularly difficult time with cymbal crashes and reverb tails, though to be fair I will admit I can't back these up with ABX studies or blind observations. Regardless, I think it's difficult to judge MP3 as a codec by iTunes 320kbps, as to be fair you'd need to consider the relative state-of-the-art in the form of LAME, or one of the newer AAC encoders.

2. Regarding the "real world" scenario with iTunes encoding, I think when you're going down the path of transparency tests in encoding you would really need to discard this constraint/assumption. If we were shooting for a "real world" test, I'd suggest using 128kbps AAC, as that's been the default iTunes encoder setting since 2003, and most likely the most common compression setting for the average listener. I don't think I'm going out on a limb when I say it would be relatively easy (though harder than most of us think) to pick out artifacts here. If an informed "audiophile" or less casual listener were actively shooting for transparency, and using iTunes, they'd probably go 256-or-greater kbps AAC (the "iTunes Plus" setting,) or would use another LAME-driven encoder and use MP3 v0-v3.
post #43 of 51
Quote:
Originally Posted by danroche View Post
Vel,

I would submit a couple of things for consideration/further research here, and this is all in good fun:

1. Reading over the comment about snares, I do believe that these might have been artifacts from using the MP3 encoder in iTunes, largely because listening-tests, even among informed listeners, haven't observed these kinds of regularly occuring artifacts in v0 and 320kbps LAME MP3s (and believe me I've personally looked for them.) I do recall, and given this was several years ago, noticing iTunes-encoded MP3s having a particularly difficult time with cymbal crashes and reverb tails, though to be fair I will admit I can't back these up with ABX studies or blind observations. Regardless, I think it's difficult to judge MP3 as a codec by iTunes 320kbps, as to be fair you'd need to consider the relative state-of-the-art in the form of LAME, or one of the newer AAC encoders.

2. Regarding the "real world" scenario with iTunes encoding, I think when you're going down the path of transparency tests in encoding you would really need to discard this constraint/assumption. If we were shooting for a "real world" test, I'd suggest using 128kbps AAC, as that's been the default iTunes encoder setting since 2003, and most likely the most common compression setting for the average listener. I don't think I'm going out on a limb when I say it would be relatively easy (though harder than most of us think) to pick out artifacts here. If an informed "audiophile" or less casual listener were actively shooting for transparency, and using iTunes, they'd probably go 256-or-greater kbps AAC (the "iTunes Plus" setting,) or would use another LAME-driven encoder and use MP3 v0-v3.
Fair points about the encoder. I am curious to try LAME/AAC now and may do so sometime soon against the original FLACs. Probably make another post as to my results.

I know we're not supposed to talk about pirated music but this is what I'm meainly meaning by "real world". Fact is, most people (at least that I know) have pirated music from various sources. These sources are unlikely to know/care about audio quality, at least not past the 'big number = good' knowledge. It is a safe assumption that many of the 320kbps downloads out there were not made at optimal settings, and in all likelihood were ripped using iTunes then shared online. So, due to this, I am trying to encourage interest in FLACs which are much harder to obtain from pirated sources and can encourage actual sales of CD's. If you think you can get something for free which is of the same quality as paying for it, there is less incentive to buy. If you know there is a high probability even a 320kbps download will not match a CD then you may be more inclined to buy it. Hopefully that makes sense
post #44 of 51
I think the difference is pretty small, but each time my assumptions where right, I think the problem with mp3 was instrument/frequency separation:

foo_abx 1.3.4 report
foobar2000 v0.9.6.9
2010/03/12 22:39:35

File A: D:\Muziekarchief\teveel versie\Arrival Project - Kazantip (Sean Tyas Remix).mp3
File B: D:\flac\Arrival & Fonarev - Kazantip (Sean Tyas Remix).flac

22:39:35 : Test started.
22:43:29 : 01/01 50.0%
22:44:27 : 02/02 25.0%
22:48:10 : 02/03 50.0%
22:49:47 : 03/04 31.3%
22:52:37 : 04/05 18.8%
22:55:30 : 04/06 34.4%
22:56:38 : 05/07 22.7%
23:06:56 : 06/08 14.5%
23:07:03 : Test finished.

----------
Total: 6/8 (14.5%)

The mp3 is illegal source, VBR track with mean bitrate 222, and frequency cutoff at around 18khz. The flac is derrived from a wav I payed 1.75 pound for.

It is an EDM, trance track, which tend to have few elements you can distinguish ripquality from, and I tested several parts of the track, intro, buildup, part before climax, climax, outro, buildup some parts twice.
and first I thought I couldn't distinguish lossless from lossy quality, so I am quite happy
post #45 of 51
For me the difference in how the cymbal crashes sound is pretty obvious. I can get a perfect score in ABX tests if i focus solely on how the cymbals sound. (And thats with my cheap earbuds, straight out of my desktop that has no hifi components to boast about)

Though i got to say besides that, i can barely pick out other differences btw 320kb and flac. Heck 192kbs and flac sound the same to me most of the time.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Sound Science
Head-Fi.org › Forums › Equipment Forums › Sound Science › 320 kbps vs FLAC?