Audio Compression Experience

[Solitaire]

At the begining of the song the guitarist opens up by playing
a few chords from a guitar with soggy strings - every note seems
oddly out of tune. The drummer, being a bit absent minded left his
drum kit out in the rain the night before, joins in - symbols shimmer
like overfull pots and drum hits splatter droplets everywhere. Sipping
on a drink, the vocalist starts singing the lyrics of the song without
bothering to swallow the liquid in her mouth first. The performance
becomes stranger still when the pianist dons scuba gear and slips
into a sea. Drifting up from the depths you hear the erie sound of
an underwater piano. Clearly he didn't make this artistic choice on
his own because the sound of a symphony orchastra mixing with
the waves surges forth out of the ocean. Throught the entire
performance you feel the need to shake your head, not to the
beat, but to the overwhelming feeling of swimmers ear...

P.S. Hoping to win the September Language Award. :)

[Mercutio]

This perfectly describes my experience listening to the only version I could find of a Tom Waits rarity "I'll never let go of your hand". The song is achingly beautiful, the listening experience heartbreakingly bad. He never released it on an album, so the only copy I know of is a badly-compressed one on the 'net.

If anyone can find a "clean" copy of that song, I'll gladly trade a kidney.

[arcticpenguin]

Some compression schemes are "lossy", others are not. The lossy ones can achieve higher compression ratios because they are throwing out some of the music. MP3 is a lossy format. Convenient for a 'walkman' perhaps, but I wouldn't want to listen to it on a good stereo setup.

[DoNotDisturb]

Originally posted by arcticpenguin
Some compression schemes are "lossy", others are not. The lossy ones can achieve higher compression ratios because they are throwing out some of the music. MP3 is a lossy format. Convenient for a 'walkman' perhaps, but I wouldn't want to listen to it on a good stereo setup.

Exactly. That is why I always encode in ogg vorbis. :)

[shanek]

Ogg vorbis is still a lossy compression algorithm; it just gives you greater compression for less loss than mp3.

[thrombus29]

Gots to get the APE.

http://www.monkeysaudio.com/

Only problem is the Ipod won't work with any of the Lossless compressions...yet.

[WildCat]

flac and shorten are very good lossless compression schemes, but the files are still huge, though 45% or so smaller than they would be otherwise.

[ZeeGerman]

I use 160 kBit/s MP3 on my iPod and it's just great to have about 2500 tracks available at your fingertips. I'm not an audiophile, so I can't hear the difference to a CD anyway.

Zee

[jj]

Originally posted by arcticpenguin
Some compression schemes are "lossy", others are not. The lossy ones can achieve higher compression ratios because they are throwing out some of the music. MP3 is a lossy format. Convenient for a 'walkman' perhaps, but I wouldn't want to listen to it on a good stereo setup.

Depends on the bit rate, Penguin.

It all depends on the bit rate and the quality of the encoder. That goes for any of the other lossy systems, too.

[gnome]

I would love to engage in some double-blind tests to determine how easy it is to distinguish between an MP3 and higher quality reproductions.

I played Peter Gabriel's "Solsbury Hill" into earphones from an MP3 player and it sent chills down my back. I have a hard time imagining that I missed out on anything.

But, I've never tested it. So, I could very well be wrong.

[jj]

Originally posted by gnome
I would love to engage in some double-blind tests to determine how easy it is to distinguish between an MP3 and higher quality reproductions.

I played Peter Gabriel's "Solsbury Hill" into earphones from an MP3 player and it sent chills down my back. I have a hard time imagining that I missed out on anything.

But, I've never tested it. So, I could very well be wrong.

Speaking as someone who's done that a few zillion times (look at my www button for more details), it all depends entirely on the bit rate and the encoder used.

[gnome]

Originally posted by jj


Speaking as someone who's done that a few zillion times (look at my www button for more details), it all depends entirely on the bit rate and the encoder used.

At what point does the improvement in bit rate become indistinguishable to human ears?

The thought originally came up reading an old snopes thread about how some audiophiles were fooled into believing that by drawing green marker around their CD's they could improve the sound. It occurred to me that the best way to really tell if one reproduction is better than another is by our old friend the double-blind test.

Well done page, by the way... what kind of results did you get from your tests? Anything interesting besides your basic conclusion that you've given us...?

[davidhorman]

If you take an audio file, compress it, decompress it, then subtract this new version from the old version, could you use the result as an objective measure of the loss due to compression?

David

[ZeeGerman]

Originally posted by gnome


At what point does the improvement in bit rate become indistinguishable to human ears?

The thought originally came up reading an old snopes thread about how some audiophiles were fooled into believing that by drawing green marker around their CD's they could improve the sound. It occurred to me that the best way to really tell if one reproduction is better than another is by our old friend the double-blind test.

Well done page, by the way... what kind of results did you get from your tests? Anything interesting besides your basic conclusion that you've given us...?


Such a test was done by c't , a very professional computer magazine some time ago (sorry, couldn't dig it up) where they took a variety of people from normal music consumer to audiophile, put them in front of very expensive hifi equipment - both speakers and headphones - and made them distinguish between mp3 and CD. They used different bit rates for conversion (as jj says, it's the bit rate that matters). If I recall correctly, basically the whole sample of people approached the 50/50 hit/miss rate at around 128 kB/s. So, if you are not an audiophile and your equipment is not top class, i.e. thousands of $ worth, then you should be just fine with 128 kB/s . I use 160 kB/s just to be sure I'm doing better than necessary.

Zee