MP3 Degradation, an experiment.

[back2onett]

Oh yeah, I just realised I answered a completely different question. Well it might still be useful for someone so I'll leave it there for now. I don't know nearly enough audio encoding to make a decent post on that so I'll leave it.

[Recessive Trait]

yay, a good topic! nice work dep.

i am softening in my stance on the mp3/wav debate, enough to try to hold my criticism of dj friends who only play 320s (not that i would ever be caught dead playing one out - but i come from the old school and learned on vinyl). i realise that vinyl and wavs are expensive. but there's a slippery slope. without addressing the issue of buying music, there are some djs who might have mostly 320s, but then there's that one track that they "couldn't find" at better than 192, and that other hot single that they could only find at 128, and suddenly you've got some shit hot local dj kid playing 128s on a 10k system and it sounds like he's running the sound through mud.

sorry for the mildly off topic rant. increasingly i see these young djs "ooh he's such a good dj" and then he's using ableton to do straight up mixing of 128 bangers. times have changed. or have they?

[Shekul]

yay, a good topic! nice work dep.

i am softening in my stance on the mp3/wav debate, enough to try to hold my criticism of dj friends who only play 320s (not that i would ever be caught dead playing one out - but i come from the old school and learned on vinyl). i realise that vinyl and wavs are expensive. but there's a slippery slope. without addressing the issue of buying music, there are some djs who might have mostly 320s, but then there's that one track that they "couldn't find" at better than 192, and that other hot single that they could only find at 128, and suddenly you've got some shit hot local dj kid playing 128s on a 10k system and it sounds like he's running the sound through mud.

sorry for the mildly off topic rant. increasingly i see these young djs "ooh he's such a good dj" and then he's using ableton to do straight up mixing of 128 bangers. times have changed. or have they?

My mates band had some local DJ support them in some crappy club up north. Forgot his name, but basically all the kids loved him there, he played all Hardcore dance stuff. What's funny is that he had no decks. No no, this kid USED ITUNES to DJ. Just had playlists with different songs. What's even more funny is that he was bobbing along like he was beatmatching and clicking next like he just mixed two tunes together like butter, with a look of proudness on his face. At least everyone was dancing like crazy i guess...

[DJ Crackle]

This thread is great. I'm always curious about exactly wtf is going on in mp3 encoding.

[stompzi]

At least everyone was dancing like crazy i guess...

Ugh, how? Or did he actually manage to sync them somewhat with just itunes?

[Shekul]

At least everyone was dancing like crazy i guess...

Ugh, how? Or did he actually manage to sync them somewhat with just itunes?

Yeah apparently theres like this Itunes party mode thing which syncs them, like some sorta fake dj mode. Oh and the fact that most people were pissed off their face helped.

http://www.apple.com/support/ilife/tuto ... it4-6.html LOL man, i'd laugh if Autechre done this or something

[DJ Crackle]

you've got some shit hot local dj kid playing 128s on a 10k system and it sounds like he's running the sound through mud.
?

A large portion of Austin Texas's younger dubstep scene described to perfection.

My mates band had some local DJ support them in some crappy club up north. Forgot his name, but basically all the kids loved him there, he played all Hardcore dance stuff. What's funny is that he had no decks. No no, this kid USED ITUNES to DJ. Just had playlists with different songs. What's even more funny is that he was bobbing along like he was beatmatching and clicking next like he just mixed two tunes together like butter, with a look of proudness on his face. At least everyone was dancing like crazy i guess...

Fucking shit. That's terrible.



I guess this is the place to ask...
What's the difference between using a "real" mp3 encoder like LAME, vs say... iTunes?
I know I've had it pointed out to me many times to use LAME, and I don't understand why. What do they do differently; what makes iTunes mp3 conversion not "real"? Is it just the "decision making" process the encoder goes through that static_cast referred to?

[EDN]

This thread gives me a music geek boner. Good work Depone.

It's not about messing with Dj Windows Media Player guys, he's big in the game believe it.

Kinda on a semi-related tangent, does anyone know how FLAC's work?
I have been pretty much downloading stuff exclusively on FLAC (and .wav and .aiff obviously) for a while now, and I genuinely have no idea how it works.
It sounds fucking awesome though!

[Sinisterbeats]

for those that want to get very in depth with mp3, look up entropy and huffman coding. Had a very in depth module when I was uni on how the mpeg stuff actually works.

[stompzi]

Kinda on a semi-related tangent, does anyone know how FLAC's work?
I have been pretty much downloading stuff exclusively on FLAC (and .wav and .aiff obviously) for a while now, and I genuinely have no idea how it works.
It sounds fucking awesome though!

How in-depth do you wanna go?

Lossless compression like FLAC basically throws away all the psychoacoustic and audio engineer stuff and says hey, we're gonna keep all the information, but store it efficiently (well, tradeoff CPU time for storage space) so it's smaller. Kinda like a zip file, but specialised for audio.

FLAC uses a bunch of different techniques, but the simplest to understand is Run Length Encoding. Digital audio (or digital anything) is just a bunch of 1's and 0's, right? Well, imagine this is your audio file:

11000000000111100000000

Lots of repetition. If you were saying that out loud, you'd probably say "two 1's, nine 0's, four 1's, eight 0's". So, store it like that:

21904180

Less space. Except of course, it'd be binary. But you get the idea. There's also a shitload of more complex techniques on top of that. (http://en.wikipedia.org/wiki/Golomb_coding)

[EDN]

Aha I see so taking your example you would have (hexidecimal for 2) 1 (hexidecimal for 9) 0 etc... kinda?

[stompzi]

Sort of yeah - I don't know exactly how the data is stored (there's a spec, somewhere), and it doesn't affect much anyway because the core of FLAC is way more than RLE - but the point I was trying to get across was that it stores the exact information that the source has, but in as efficient a way as possible while still allowing it to be recreated exactly.

[paravrais]

Kinda on a semi-related tangent, does anyone know how FLAC's work?
I have been pretty much downloading stuff exclusively on FLAC (and .wav and .aiff obviously) for a while now, and I genuinely have no idea how it works.
It sounds fucking awesome though!

How in-depth do you wanna go?

Lossless compression like FLAC basically throws away all the psychoacoustic and audio engineer stuff and says hey, we're gonna keep all the information, but store it efficiently (well, tradeoff CPU time for storage space) so it's smaller. Kinda like a zip file, but specialised for audio.

FLAC uses a bunch of different techniques, but the simplest to understand is Run Length Encoding. Digital audio (or digital anything) is just a bunch of 1's and 0's, right? Well, imagine this is your audio file:

11000000000111100000000

Lots of repetition. If you were saying that out loud, you'd probably say "two 1's, nine 0's, four 1's, eight 0's". So, store it like that:

21904180

Less space. Except of course, it'd be binary. But you get the idea. There's also a shitload of more complex techniques on top of that. (http://en.wikipedia.org/wiki/Golomb_coding)

I think the fact that the first thing I thought after reading that was "hey, that's really cool" is a testament to how much of a geek I am XD

[EDN]

I think the fact that the first thing I thought after reading that was "hey, that's really cool" is a testament to how much of a geek I am XD

Cheers Stompzi!

[nowaysj]

Again, the point is that these artifacts sound "nice", whereas mp3 artifacts don't.

Anything to do with "nice" is always subjective. Just heard for the nth time that in a blind test, younger people prefer the sound of mp3 (and mp3 compression) over other media, ie vinyl, cassette, and cd.

Sounds crazy right? But whatever you get used to, that's what you like. I've been there.

There was a time early in mp3 when I was exclusively listening to 128 -192, and did develop a taste for the high freq crunch.
==

Side note, congrats on the vinyl release, 110% about time.

[staticcast]

Again, the point is that these artifacts sound "nice", whereas mp3 artifacts don't.

Anything to do with "nice" is always subjective. Just heard for the nth time that in a blind test, younger people prefer the sound of mp3 (and mp3 compression) over other media, ie vinyl, cassette, and cd.

Sounds crazy right? But whatever you get used to, that's what you like. I've been there.

There was a time early in mp3 when I was exclusively listening to 128 -192, and did develop a taste for the high freq crunch.
==

Yeah, I know what you mean. The transient smearing almost sounds a bit like an aural exciter...

Side note, congrats on the vinyl release, 110% about time.

Cheers

[staticcast]

Kinda on a semi-related tangent, does anyone know how FLAC's work?
I have been pretty much downloading stuff exclusively on FLAC (and .wav and .aiff obviously) for a while now, and I genuinely have no idea how it works.
It sounds fucking awesome though!

How in-depth do you wanna go?

Lossless compression like FLAC basically throws away all the psychoacoustic and audio engineer stuff and says hey, we're gonna keep all the information, but store it efficiently (well, tradeoff CPU time for storage space) so it's smaller. Kinda like a zip file, but specialised for audio.

FLAC uses a bunch of different techniques, but the simplest to understand is Run Length Encoding. Digital audio (or digital anything) is just a bunch of 1's and 0's, right? Well, imagine this is your audio file:

11000000000111100000000

Lots of repetition. If you were saying that out loud, you'd probably say "two 1's, nine 0's, four 1's, eight 0's". So, store it like that:

21904180

Less space. Except of course, it'd be binary. But you get the idea. There's also a shitload of more complex techniques on top of that. (http://en.wikipedia.org/wiki/Golomb_coding)

There's some more detail here:
http://flac.sourceforge.net/documentati ... rview.html

It's pretty clever stuff -- instead of storing all the data, they store an approximation function (like a polynomial or something, I'm not exactly sure) and then the residual error, since the residual error requires much fewer bits to store. If you evaluate the function and add the error, you get the original value.

[Depone]

Bump for the people who missed this

[Jak The lad]

Nice read, think I was mainly drawn in by the colours lol.

[graish]

Sometimes when I upload tracks to soundcloud I can hear a sort of ringing in the sub bass, anyone else know what I'm talking about?

The low end in these examples similar, even down to the lowest bitrate.