audio questions

[Bouncing Zabaglione Bros.]

Good luck!

To be fair, you didn't choose music that was going to show much difference. Chose anything with heavy synths and big layered sound (trance, Vangelis, Human League, etc) and compression differences do become noticable, but I think that's down to a squeezing of the dynamic range. Big sweeping instruments become less bright and more constrained and gain a sort of "listening down a metal tube" element at higher compression.

One reason why I prefer Ogg is that it attempts to hit sound quality targets in preference to bitrate targets as MP3 does. Ogg does seem to be better at that kind of "problem music" than MP3 for any given filesize/bitrate.

Having said that, modern albums with their over-amped, over-clipped production and it's associated loss of dynamic range is a much worse and more noticeable degradation in sound quality IMO.

 

[Albuquerque]

To be fair, you didn't choose music that was going to show much difference. Chose anything with heavy synths and big layered sound (trance, Vangelis, Human League, etc) and compression differences do become noticable, but I think that's down to a squeezing of the dynamic range.

Oh now BZB, while an intelligent person might be able to agree with you, please instead consider the true context of this thread. Let me "enlighten" you a bit:

I don't know how many times i have to say this, but even though i have lousy speakers, I DO notice a great magnitude of difference between an mp3 and the same track in wma lossless. I'd like to know why people keep on telling me i can't notice something when I actually do. When people try to tell me that I can't notice a difference when I actually can, for no good reason, it kind of pisses me off. Please don't make some completely retarded incorrect assumption about me.

I mean after all, wma lossless wouldn't exist if the average music listener like me couldn't tell the difference on $50k stereos, when in fact people with $20 headphones can probably tell the difference between an mp3 on myspace and the same track downloaded from music giants, or a cd track ripped in mp 3 vs wma lossless.

He hears a great magnitude of difference between an MP3 and the same track in WMA lossless. He didn't specify what kind of sound, and to be just as fair, that's his own damned problem. So according to his own words, he will hear the absolutely massive difference between the absolutely shit-tastic lossy formats I've chosen and the pure PCM audio rip of the CD.

Which is why he will be able to directly pick out the audio stream that was never compressed to begin with... Right?

 

[Simon F]

I think this thread has run its course.... or nearly run its course....

 

[Albuquerque]

Thank you Simon.

If he doesn't reply back in ~48 hours, I'd certainly agree that it's done and we can assume he just isn't able to tell the difference. And then I can pull those samples from my wifes' webspace

 

[digitalwanderer]

This thread took a sharp turn from boring to pretty awesome.

Very much so, brilliant play Albuquerque!

 

[Albuquerque]

Any of the long-time regulars whom I know are welcome to PM me which ones they think are which, if they want to test their skills I will eventually post the answers in this thread, unless of course 2008 just nails it on the first try.

 

[vazel]

For reference here is a mp3 vs. CD test. http://www.geocities.com/altbinariessoundsmusicclassical/mp3test.html

 

[L233]

For reference here is a mp3 vs. CD test. http://www.geocities.com/altbinariessoundsmusicclassical/mp3test.html

Ah, the famous c't test which essentially established that you have to be hearing-impaired with a gimp ear in order to hear a difference between a 256 kpbs MP3 and a CD (because then the psycho-accoustic model may fail you).

That's the beauty of blind tests.

 

[Albuquerque]

Fantastic article Vazel, I hadn't seen that before. This paragraph from the "summing up" section highlights the most interesting point for me:

The fact that some of the 128 kbps samples were consistently judged to be better than their original CD counterparts by this skilled group – even by the best among them – stunned our editor (who participated in the test although his results were not included in the evaluation, and had to confess that he got only 15 points). It seems safe to declare that there is no musical genre that is especially well-suited or ill-suited to compression. It is apparent that there are quite other factors related to the technical aspects of recording that will later adversely affect the results at low bit rates.

I know we've discussed the relative merits of various source media versus it's compressed result, but their analysis generally showed that there's not any sort of "general rule" as to what genre of music will show more deterioration. I find it intriguing (and yet, not surprising, based on what MP3 encoding is supposed to do) that so many of these highly skilled people actually chose the MP3's as the better sounding format over the raw CD output.

Another interesting note in the article: MP3's (again, by virtue of the fundamental way they decide to compress the music) can effectively hide some of the mastering flaws that might exist in the raw media. Actually giving further credence to the compressed version sounding truly better than the uncompressed version...

 

[2008 IQ is unacceptable]

lepno sounded different. not a huge difference, but it was the black sheep of the 5.

Dynamic range is definitely hurt by compression; that's the most noticeble effect of compression to me.

Besides, some songs are hurt more by compression than others. I can tell a huge difference between the myspace version of let go, by frou frou (imogene heap) and the wma lossless version i downloaded from musicgiants.

Another song i notice a huge difference in is paramore's "that's what you get."

I'll grant you that it wasn't incredibly easy to tell the difference between the 5 samples you gave me, but it always varies based upon the track.

ymsig sounds the best, imo.

 

[Albuquerque]

lepno sounded different. not a huge difference, but it was the black sheep of the 5.

Dynamic range is definitely hurt by compression; that's the most noticeble effect of compression to me.

Besides, some songs are hurt more by compression than others. I can tell a huge difference between the myspace version of let go, by frou frou (imogene heap) and the wma lossless version i downloaded from musicgiants.

Another song i notice a huge difference in is paramore's "that's what you get."

I'll grant you that it wasn't incredibly easy to tell the difference between the 5 samples you gave me, but it always varies based upon the track.

ymsig sounds the best, imo.

Quoted for posterity Now we'll wait for Scott_Arm or Rys (they both have the answers; wasn't sure if Scott was originally going to accept my volunteer effort)

I will make one statement for now: in a single post, you've gone from a rock solid "I can tell the OBVIOUS difference and lossy compression is shite" to a cushy "Well, it kinda depends you see..." Collectively, this forum has been telling you this all along, and until now, you've held hard and fast that it doesn't matter and that you can always tell.

Keep this in mind when the results come back...

 

[Bambers]

hmm hard test being that piano tends to compress well, answers (guesses) pmed

 

[Rys]

chmjv is uncompressed, lepno was nowhere near being the black sheep (axkfa was the most compressed), and your favourite was WMA.

I couldn't tell without spectral analysis, personally, which was the point.

I respectfully request that your one man crusade against data compression now take a back seat to more fruitful pursuits.

Excellent test, Albuquerque, great stuff.

 

[digitalwanderer]

So I shouldn't feel dumb for not being able to tell them apart then? :|

 

[2008 IQ is unacceptable]

chmjv is uncompressed, lepno was nowhere near being the black sheep (axkfa was the most compressed), and your favourite was WMA.

I couldn't tell without spectral analysis, personally, which was the point.

I respectfully request that your one man crusade against data compression now take a back seat to more fruitful pursuits.

Excellent test, Albuquerque, great stuff.

I thought axkfa and zqrtf sounded the worst. Between the other 3, I couldn't tell which was the best, before it was pointed out.

I'll end my crusade, but it does depend on the track and the algorithm used, and whether or not it's a blind test. Those 3 are rather large factors.

 

[Thowllly]

I thought they all sounded exactly alike, except axkfa which was different, it had a rougher sound. But I wasn't able to determine if it was better or worse than the others, so I was sure it was either the 128kbps or the uncompressed one, but I didn't know which one!

-One one hand I found it unlikely that axkfa was the uncompressed one, because how could all the different compressions transform that rough sound exactly the same way into exactly the same smooth sound?

-On the other hand I thought axkfa sounded better than the others because it sounded like my mothers old creaky untuned piano (although it was more tiresome to listen to than the others....)

I identified lepno.wav as the 500kb/s one, but that was only from looking at the file, not listening to it. As I said, with the exception of axkfa they all sounded exactly the same to me (only have crappy headsets in my defense).

 

[L233]

I burned the tracks to CD. My stereo certainly isn't audiophile level but while it's old, it's quite a notch above what most people have (Sony TA-F 550 ES amplifier, Technics SL-PS 700 CD player and JBL LX 500 speakers). In addition, I used my Sennheiser HD 595 headphones.

I identified "axkfa" as the most compressed but it wasn't nearly as easy as I thought it would be, it was basically just some artifacts here and there that gave it away, rather than the track sounding consistently worse.

The rest of them.... I can't tell them apart, at least not to the extend at which I can be confident that I heard a difference rather than just imagined one. At any rate, I certainly could not say which one sounds better or which one is the uncompressed.

 

[Bambers]

Woohoo i guessed (the uncompressed and most compressed) right

Order I PMed to Albuquerque was:

axkfa
ysmig
zqrtf
lepno
chmjv

Axkfa was pretty obviously quite poor but that was the only one that stood out as being 'wrong'

Lepno sounded 'cleaner' than the other 3 remaning ones but, depending on the original recording that might have been a good or bad thing.

Thats really about all I could tell from normal listening mind.

Then I fiddled with the mixer and maxed the xifis 24bit crystalizer (which has a habit of bringing out compression problems since it boosts the really high stuff) and the 8k and 16k on the EQ and concentrated on the first couple of bars and deduced that:

lepno did indeed have less 'noise', ymsig and zqrtf sounded essentially identical and chmjv was different again. By logic this leaves only lepno or chmjv as the uncompressed sample and I went for chmjv simply because lepno sounded just too clean, presumably the compression had ditched the essentially inaudable recording noise. The remainig three i just stuck in a random order.

Eitherway though, aside from the 128kbit which really is quite poor and maybe lepno sounding a little 'too' clean to be true, it's impossible to pick out the 'right/best' sample without doing silly stuff to your mixer which isn't exactly normal listening conditions.


However I did notice that one of the strings on the note played on at the beginning of the 24th and during the 27th and 30th seconds is out of tune

 

[Scott_Arm]

Rys beat me to it I've been too busy at work.

 

[Albuquerque]

The WMA was actually my favorite too of the bunch...

I'll end my crusade, but it does depend on the track and the algorithm used, and whether or not it's a blind test. Those 3 are rather large factors.

So let's recap: your favorite was WMA audio. The lossy compression rate on that file was slightly over 9:1 -- a 7880kb file was shrunken down to about 840kb. We now have room for nine times more audio with the same perceptual quality when comparing to uncompressed PCM.

And the part about whether or not it's a "blind test" is irrelevant -- the only reason it sounds different when you know the format is because your mind tells you that it needs to sound different. If I had purposely mislabeled every single one and told you that all the labels were wrong, it probably would've been even harder for people to cipher which ones were which. I goofed a small amount on the OGG as it came out as a 48khz sample rate in the PCM, which gave away it's true nature for those who knew to look.

Now for your first point regarding the source audio: for what it's worth, the track I selected was a bit biased -- any recording of an "analog piano" (ie, the real thing and not synthesized) is almost always one of the noisiest recordings due to the all the strings and vibration and resonance from the chassis. This effectively makes compression even harder to spot...

If anyone is interested, I'll gladly do the same set of compression algorithms on an "HDR" audio track -- I can pull some music off my Time Warp disc that I mentioned earlier in this thread. I'd be willing to wager that we end up in the same scenario, or at least a very similar one. And we could be less secritive about the end results I could just give everyone a day or so to tally up their ideas, converse about it, etc...