You are not logged in. Please login or register.
Robots for Robots → The Daily UFO → empirical fact: good mp3 sounds as good as wav.
Yes, my player supports ogg, thats nice.
This topic is very interesting...I'm not very familiar with open-source codecs but I think I will rip my cds and vinyls with flac now.
Shit, I need a bigger HDD....
I have studied a little how lossy codecs work, and this time I did the test with a sound that is supposed to put the algorithm in a difficult situation. And indeed, this time, the difference is audible at any level of compression.
I also based the new test on the idea that mp3's sound flatter in a night club, they seem to miss energy, like "bass have no balls and treble are so-so". So I exposed the algorithms to a wide frequency-range sound with fast straight attack modulation. Indeed, here, lossy codecs are pulling the tongue...
The MP3 lame codec
Audio version of the test (original sound was pretty tough for the mp3 algorithm... at 320k the difference is already audible, it still tries its best at 192k but below... game over. At 128k it's catastrophic)
http://www.fileden.com/files/2006/11/14 … st-mp3.wav
The WMA 9.2 Codec
I don't like it, but objectively, it muddles through pretty okay at 192 kbps...but with that sound it abandons and become crappy at 128 k
Audio version:
http://www.fileden.com/files/2006/11/14 … st-wma.wav
The Ogg Vorbis AoTuv SSE-3
Does the job, especially good between 128 and 320 kbps, in comparison with the others
Audio version:
http://www.fileden.com/files/2006/11/14 … st-ogg.wav
The AAC Nero codec
Better than the MP3 but not better than OGG or WMA.
It tries recreate some frequencies at low bitrate (64 kbps) but it's not the original ones, of course
Audio version:
http://www.fileden.com/files/2006/11/14 … st-aac.wav
@robotworks, sounds like your lame encoder is an old version or some other problem. i encoded this with the version 3.97 with "lame -q 0 -b 320 .."
http://www.punainen.org/~biotek/test_q0_320k.mp3
still sounds passable with "lame -b 192.."
http://www.punainen.org/~biotek/test_192k.mp3
edit: nevermind, didn't get that there were two test signals.
i did some testing of the differences between 320k lame mp3 and 320k aotuv ogg, in particular the encoding error (the stuff that you don't hear.)
original target wav with the lossy versions:
http://www.punainen.org/~biotek/test/test1.wav
http://www.punainen.org/~biotek/test/test1.ogg
http://www.punainen.org/~biotek/test/test1.mp3
resulting encoding error:
http://www.punainen.org/~biotek/test/te … eerror.wav
http://www.punainen.org/~biotek/test/test1_oggerror.wav
the results were produced with lame v3.98 and oggenc v2.85 (aoTuVb5.5). error signals were summed together in phase from the original wav and inverted versions of the decoded wavs (produced using lame and oggdec with heaviest noise shaping).
i think the results are comparable, both results pretty much only have noise (peak from the snares at around -30dB) from the transients and noise to begin with.
anyway these tests don't really measure the performance of the lossy codec since it's subjective. if you want to be totally serious about measuring it then you have to resort to statistical double blind listening tests with lots of people. these sort of tests have been done and the result is that most people don't hear the difference.
The LAME encoder I used is well the v3.97.
Be careful, in the wav files I posted, the same 2 sec. test sound is repeated 7 times, the first 2 sec is the original wav, the next 2 sec. is the 320k, next the 256, next 192, next 128 etc, with of course a total lenght of 14 sec.
So your 320k and 192k files just do sound like it sounds on the 2 to 4 sec and 6 to 8 sec segment of my file.
Regarding the file I posted for the mp3 test, you might have notived it's only the 6 last seconds that sound really crap (2 sec at 128k, 2 sec at 96k and 2 sec at 64k).
In clubs mp3 are easily spottable, because of the high volume level and because it requires more dynamics. So the test sound is a sound that exposes the algorithm to many very brutal waveform changes, in order to spot the difficulties codecs can have responding to big sudden dynamic changes. The "punchiness" of sound depends on the ability to reproduce very fast time-attack and time-decay when the waveform changes, this requires a lot of bits and tolerates hardly the compression.
Normally the mp3 codec has a "bit reservoir" that allows it to manage with such brutal changes. I takes benefit of a "hole"
in the spectrum to store them. But unfortunately I didn't let any hole in the spectrum and brutal changes in the waveform occur too often, so that "bit reservoir" is overloaded almost at once: codec don't manage with it at all at 128k, and the difference is already audible at 320k, and like you said, at 192k it is... well... "passable".
At 128 k anyway, the difference seems too obvious here to require a double-blind study... And it's a current bitrate over the web...
i don't see how this old 'club pa' argument holds any water. with loudspeakers in a room you're getting lots of reflections which add to the masking.
has flac got anything to do with fap?
i'll get me coat..
In clubs mp3 are easily spottable, because of the high volume level and because it requires more dynamics. So the test sound is a sound that exposes the algorithm to many very brutal waveform changes, in order to spot the difficulties codecs can have responding to big sudden dynamic changes.
for sure these tests don't measure a club pa situation. you overlooked the fact that the mp3 encoding doesn't have to account for the amplifying of the signal.. pa system takes the line level mp3 signal and amplifies it. just like a needle of a turntable isn't pushing the speakers.
i don't see how this old 'club pa' argument holds any water. with loudspeakers in a room you're getting lots of reflections which add to the masking.
You never heard things like "Hola, the DJ... sound is weak... I'm sure he plays mp3s"? Me yes, in that conditions, people notice it... But for home listening, the difference is much less spotable.
Robotworks wrote:In clubs mp3 are easily spottable, because of the high volume level and because it requires more dynamics. So the test sound is a sound that exposes the algorithm to many very brutal waveform changes, in order to spot the difficulties codecs can have responding to big sudden dynamic changes.
for sure these tests don't measure a club pa situation. you overlooked the fact that the mp3 encoding doesn't have to account for the amplifying of the signal.. pa system takes the line level mp3 signal and amplifies it. just like a needle of a turntable isn't pushing the speakers.
Yes but they amplify attacks and decays that are slower or less precise than a CD or vinyl. MP3 is a good thing, gladly we have a simple way to download all these songs. But is has limitations, and the data lost in the compression... well that data is lost, no doubt.
that in itself doesn't really say anything. maybe the dj isn't playing a mp3, maybe he is. maybe he is playing a crappy encoding. who knows. club setting is hardly a place for objective measurements.
There are certainly ways to increase the MP3 rendering anyway. I thought to something like frequency mirroring. It would resynth some frequencies and give more clarity to the sound...
biotek wrote:i don't see how this old 'club pa' argument holds any water. with loudspeakers in a room you're getting lots of reflections which add to the masking.
You never heard things like "Hola, the DJ... sound is weak... I'm sure he plays mp3s"? Me yes, in that conditions, people notice it... But for home listening, the difference is much less spotable.
biotek wrote:Robotworks wrote:In clubs mp3 are easily spottable, because of the high volume level and because it requires more dynamics. So the test sound is a sound that exposes the algorithm to many very brutal waveform changes, in order to spot the difficulties codecs can have responding to big sudden dynamic changes.
for sure these tests don't measure a club pa situation. you overlooked the fact that the mp3 encoding doesn't have to account for the amplifying of the signal.. pa system takes the line level mp3 signal and amplifies it. just like a needle of a turntable isn't pushing the speakers.
Yes but they amplify attacks and decays that are slower or less precise than a CD or vinyl. MP3 is a good thing, gladly we have a simple way to download all these songs. But is has limitations, and the data lost in the compression... well that data is lost, no doubt.
yes pa systems amplify music. there probably is some nonlinear processing involved like limiting and clipping which add elements to the signal but that is after the mp3 decoding in the signal path.
if you consider my test, it shows that the data lost is pretty much noise to begin with which doesn't have much information in it.
When I listen to your test, I don't only hear the continuous background noise. We recognize many elements that belong to the song itself... Remember that when you have have straight attack time on a sound, that attack covers the full noise spectrum at the begiining of the sound, it's precisely what make it more "punchy" and what is removed here. (When you have a vertical signal in a wave file, it sounds like a noisy click. it's removed by the codec so the attack stops being vertical, so you lose precision). And some frequencies corresponding to the "tone" are also present your error exctraction files....
Anyway at 320k, MP3 are good sounding, the loss is not so significant. But this high bitrate is not yet a standard over the internet...
Mp3 and other codecs are a good invention that allow many things and improve our daily life, but non-destructive sound data storing stay better for a professional use, in my humble opinion.
it seems a little perverse to me to compress vinyl rips with flac..
edit: 256k and 320k are pretty much the standard on mp3 shops.
It depends on what you intend doing with your vinyl rips. For sound sharing, FLAC has no real interest (too big files). For Public Adress, or other professional purpose you can just leave them as wav, if you needn't to send them over the internet...
50MB is too big in 2008? 
use wavs and aifs they are oldschool and real
When I download 50 MB audio files, they are .wav...
bah... I can wait 8 seconds more
FLEMMING DALUM - Cybernetic B-Boy.wav
70MB .wav = 51.8MB .zip (verified with 7zip)
Yes, I recognize 
Why would anyone zip a wav? flac them! is the "same" as zip but still working 
just buy vinyl, then you can listen to music instead of flaccing, zipping, encoding, decoding, compressing and endlessly debating about it.. 
listening to her new clone.nl order
Robots for Robots → The Daily UFO → empirical fact: good mp3 sounds as good as wav.
Powered by PunBB, supported by Informer Technologies, Inc.
Currently used extensions: pun_approval, pun_admin_add_user. Copyright © 2009 PunBB