Slashdot Mirror
The Death of High Fidelity
Ponca City, We Love You writes "Rolling Stone has an interesting story on how record producers alter the way they mix albums to compensate for the limitations of MP3 sound. Much of the information left out during MP3 compression is at the very high and low ends, which is why some MP3s sound flat. Without enough low end, 'you don't get the punch anymore. It decreases the punch of the kick drum and how the speaker gets pushed when the guitarist plays a power chord.' The inner ear automatically compresses blasts of high volume to protect itself, so we associate compression with loudness. After a few minutes, constant loudness grows fatiguing to the brain. Though few listeners realize this consciously, many feel an urge to skip to another song."
I call shenanigans. Double blind testing shows no perceptible difference between a good MP3 and the source material for most listeners most of the time. The real death of hi-fi is the fault of the record companies themselves, and the Loudness War. Who cares if an MP3 encoder drops a tiny amount of imperceptible data when the CD itself has been compressed and clipped to the point that you don't want to listen to it?
People who think MP3 encoded with Lame -preset standard (about 192kbps) suck and are not trolling should register at Hydrogenaudio and submit audio samples and ABX tests tests. Some Lame developers hang out there, and I'm sure they would like some help in improving their acoustic model.
The problem is that the waveforms of modern songs are increasingly rendered at a uniform loudness, causing listener fatigue (it sure makes me tired). This is well addressed in the article.
MP3 compression is yet another issue.
I think you're just getting older!
Reality is defined by the maddest person in the room
I think you resumed in two sentences the whole "audiophile" dilemma. Let's face it, modern recordings suck and no processing will change that. Meanwhile, well intentioned but ill informed people will debate endlessly if vacuum tubes are better than transistors, if analog is better than digital, if lossless compression is better than lossy.
Raising these subjects is flamebait, the people who defend vacuum tubes or analog recordings are comparing their own favorite recordings with modern recordings, not the absolute value of the audio equipment itself.
One of my own favorite musics is a recording of the nine Beethoven symphonies, done by the Berliner Philharmoniker, conducted by Herbert von Karajan in 1962-1963. I have several versions of these in both analog medium, tape and LPs, and also in CDs, which I have ripped to mp3 to carry in my portable player. To rip the mp3 I used the CDs, not any of the analog versions, because the sound is cleaner in the CDs.
OTOH, I have also some other CDs of those same pieces, same orchestra, same conductor, same recording company, done entirely in digital formats. I think they aren't as good as the old ones. The reason? Not because they are digital, but because of the difference between a Karajan in his 30s compared to the same man 20+ years later. Or it could also show the difference between the criteria used by Deutsche Gramophon in the 1960s and the 1980s.
However, one thing I'm sure of is that if a CD copy of an analog recording is better than an analog copy of the same recording you cannot say digital sound is inferior. And if an mp3 copy of a CD containing music originally recorded in analog format sounds better than an LP of exactly the same recording, you cannot say mp3 has intrinsic fidelity problems.
The loudness wars have been going on with commercial radio for quite some time. See the infamous Optimod or Omnia. One of the tenants of processing is to make younger audience music squashed to death (heavy overdrive and heavy clipping) because they apparently don't care about fatigue.....but to a middle-aged soccer mom--the typical targeted demo of the greater majority of stations--the processing gets very fatiguing so they just clip it to death without the massive overdrive, still causing horrible distortion.
Next time you have the radio on, listen closely...those little crackles in the background is not noise from a bummy signal, it's distortion from over-processing the already over-processed song.
Music that's older (recorded when the technology wasn't so hot) comes pre-clipped because they didn't have amazing compression devices to keep everything in check so the varying levels max out. It's not as bad since it were tubes causing the clipping (and they have a softer sound), but it sounds awful.
Anonymous because this is my profession.
...remove anything at the bottom end of the spectrum. There is simply no point as the entire low frequency range can be represented by just a few coefficients.
The authors have no idea what they are talking about and are probably a combination of prejudiced and stone deaf.
You can still hear most of the dynamic range on a well encoded MP3 or Vorbis file, IMHO. If it's present in the first place, that is.
Never mind discussing whether FLAC or MP3 or OGG are the best ; what does it matter if the master has already been sabotaged by marketing, compressed to sound "loud" so that it gets instant attention on the radio? Yeah, sure, it gets attention ; the same way a fire alarm or a fog horn does, by inflicting an ear-cringing reflex. "Compression is a necessary evil. The artists I know want to sound competitive. You don't want your track to sound quieter or wimpier by comparison. We've raised the bar and you can't really step back."
-- Butch Vig, producer and Garbage mastermind Yes, this man truly is a mastermind
Much of the information left out during MP3 compression is at the very high and low ends, which is why some MP3s sound flat.
Wait, I thought that the MP3 compression was basically achieved by cutting the sound into overlapping chunks, performing a DCT on each chunk, discarding the less important bins according to a psychoacoustic model and compression the thing like in a ZIP file? If so that means that the frequency scale stays linear, and so there would be little interest in getting rid of frequencies under say 30-35 Hz since they represent about 0.15% of the data in a plain old track sampled at 44,100 Hz.
So the MP3 compression doesn't actually discard the "low end" as they call it, does it? Wouldn't the "flatness" they're talking about be due to how frame sizes affect transient (short) sounds and makes them softer?
You just got troll'd!
YouTube: The Loudness War
Taking stuff apart since 1969 (TM)
that your equipment doesn't have wooden knobs.
Also, you'll find your aural experience greatly improved if the wires are of high quality and raised slightly above floor level. I've also noticed marked improvements if you chill the wires(and generally keep the room cool). Cool equipment = warm sound. Who knew?
It's called the auralgasm setup for a reason!
You need to get some of those speaker baffles made from oxygen-free copper.
It's true I tell you, feller at work's next door neighbour read it in the paper.
Well, to be fair the article is specifically talking about the phenomenon known as 'finalizing', which is a way to clearly boost the
apparent levels by up to 10 dB or more during the mastering stages without any digital clipping artifacts. (a.k.a. brick-wall limiting)
There is no question that a lot of great points were raised in the article, however when it comes to MP3 (the 'other' form of compression)
as a person who has participated in recording, mixing and mastering sessions for over 30 years, and constantly listens to master recordings,
can only say that it is pathetic how bad they sound on large audio playback systems, which some of us have and listen to.
(For example pick a very large loft, or someone's home theater for 20 people, not to say anything of a proper auditorium)
You might not hear it at home, on computer speakers or certainly not your earbuds, but the bigger the stereo, the more it is obvious.
And actually what is the most disturbing is that what is very, very wrong about lossy encoding formats is that it doesn't necessarily affect so
much the frequency response, as it does the 'punch', transients and other intangibles which when played on those large-format systems become
quickly apparent. The same way a graphic designer will not try and magnify this site's jpg logo (415 x 55 pixels, I did check) to a more
adequate 16,000 x 2122 for billboard and poster printing, as there will be obvious and nasty pixelization artifacts, there are similar phenomenons
happening with audio, and they are - at best - poorly understood, and at worst dismissed as being the brainchild of crackpots with too
much time on their hands, the New-Age idealists like those who read John Diamond's "Life Energy In Music" and keep a stack of copies
of 'Absolute Sound' by the bathroom stall.
Suffice to say that the combination of both forms of compression (finalizing, plus lossy encoding) do make for a pretty formidable opponent that
already has greatly affected the public's perception of what 'sounds good' and doesn't. And it's not likely to get better.
Fear not, for those who care about listening to music in more proper manners, there are plenty of options available, from an arguably limited selection
of SACDs of some great Jazz, Classical and Pop, to fantastic vinyl playback systems, or ways to re-process those CDs that are too loud and give them
back some form of dynamic range, which will involve spending time re-mastering them with specific analog//tube//tape-machine type equipment, and is
obviously not a recommended activity for what seems to make the most of today's impatient 'click-click' listeners, the Attention-Deficit-Disorder-addled set.
As for the Hydrogen Audio bunch that keeps doing those double-blind tests and play with oscilloscope and frequency analyzers, I think they should
once try them again, but in a place that holds a couple of thousand listeners, and they may come back around to the fact that even CD-resolution
is quite atrocious to listen to, when compared to something like formats that can actually reproduce the original master recordings in a way they should,
such as DSD or 24-bit / 96 kHz encoded music. (not to say anything of a proper 1/2" open-reel master copy)
So in essence, while some of these people quoted in the article all agree that something's wrong, most of them cannot put their finger on it, as it is
something that is far more in the domain of the perceptual and psychoacoustics than an exact science.
It is mind-boggling that 25 years after the CD was introduced, most people consider progress to be size-reduction and loudness, and all attempt
at making a case for higher-fidelity have commercially failed, but again there are far larger problems looming over our heads today.
As someone who has made a living with playing recorded sounds in very large venues, I can however vouch for the fact that even if people do not exa
I do encode my mp3s using LAME at 192 kbps and even though I would not characterize the sound as sucky, I could detect a difference between the mp3s and the original (CD played on a 13 year old relatively higher end Sony CD Player). The article is on the mark, the bass and the punch of drums at the bottom end is not as strong. I do not detect differences on the high end, perhaps because of my aging ears.
It could be that the mp3s encoded in the latest version of LAME could have closed the gap but it is also likely that the difference is exacerbated by the fact that I am playing the mp3s via the laptop's headphone jack hooked up to the stereo amp. I wish someone would manufacture an mp3 player with better analog output circuitry designed not for headphone / earphone listening but for hooking up to hifi components.
If you're going to compare CD with mp3, compare the original wav files to the mp3 instead of comparing your mp3 player to your CD player. As it is, you have too many variables. I wouldn't be surprised if there was an audible difference between a headphone jack and a line out, simply because they have to drive very different loads.
Translation of parent:
In tests, MP3s made with LAME at the default settings are usually very hard to distinguish from the original. The test is to play the original (A), then the MP3 (B) and then a random choice of the original or the MP3 (X). The listener then has to guess if X was the original or the MP3. This is repeated several times until the results are statistically valid. In most cased people, even audiophiles with high end equipment, cannot accurately determine which one X is.
const int one = 65536; (Silvermoon, Texture.cs)
SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
I just laugh at the LP's were better crowd when reading how guys like Phil Ramone were compressing the hell out music to FIT IT IN THE LP's LIMITS back then. When CD's came out he (producer of Sinatra, Streisand. Simon, Billy Joel, Ray Charles, etc etc) couldn't believe how much better the digital format was. Didn't have to compensate for needle momentum on inside tracks any more, true dynamic range and so on. Read about it in his "Making Records" book. The sound was different for LP's and we could if we want reproduce that digitally, but we don't.
That's not how bit depth is used in audio recording/playback.
Bits in audio are all about dynamic range.. you still need all the bits for loud music as well as quiet music.
16 bits gives you 96 dB, and 24 bits gives you 144 dB. This is why 16bit is "good enough" for most music, but recording is almost always done at 24 bits to allow for more accuracy of level adjustments and mixing. Then down-mixed to 16 bits.
If you really want to be 100% fair, rip the original CD to WAV (or FLAC), then reburn it. Then encode those WAVs (or FLACs) as MP3s, then decode them again, and burn that.
You can now play both on the same relatively high-end CD player. (Or you could try playing both from a laptop, if you like, but I'll bet the CD player is better.)
Don't thank God, thank a doctor!
After reading the article in Rolling Stone (several weeks ago) I came to realize that the quoted music producers didn't know the difference between the two audio definitions of word 'compression'. They were using the two different meanings interchangeably to make arguments that reflected their financial positions in the music industry rather than make sense to the music consuming public.
...And they are supposed to be professionals!
Audio compression means to reduce the amount of difference between the loudest and softest sounds of an audio recording or signal. This is what a guitar stompbox pedal like the MXR Dyna-comp does or what the NE571 Compandor IC does.
File compression is to transform the time-domain voltage samples of a digital audio recording, convert them in frequency domain, and discard data below a certain threshold.
Compression means to make smaller. Audio compression reduces volume range and file compression reduces file data size. But they are completely different concepts.
Both types of compression are done on audio recordings by the music industry. Both affect the resultant product.
But they are completely different processes that affect the music in completely different ways. And many of the music professionals quoted in the article couldn't tell or honestly didn't know the difference.
16 bits is enough dynamic range for playback, though. The CD format wasn't chosen at random: it exceeds the fidelity of the human ear. The scientists and engineers who delevoped the CD format weren't settling for "good enough". Those who say different are selling something (usually extremely overpriced audiophile gear).
For mastering and mixing of course you need more bits, so that you preserve 16 data-ful bits at the end of the process.
24 bit CDs would do *nothing* to preserve sound quality *after* dynamic range compression. The data has already been lost, adding more 0s doesn't get you anything.
More bits on the master recording might help, but that has nothing to do with the CD format, and everything to do with the mastering process.
Socialism: a lie told by totalitarians and believed by fools.
No, it isn't. It's the smartest possible choice. There is no loss of stereo separation in LAME "joint stereo" (actually, mid/side or matrix stereo), unlike in intensity stereo encoding, which isn't even implemented in LAME. How LAME works by default is that it analyses each frame separately to see whether it is more efficient to encode the frame in LR or MS. Most of the time, not every frame is encoded in "joint stereo". If there was an audible effect to stereo imaging from using MS encoding, the stereo image would continuously pump back and forth as the encoding method changes. Never heard of anyone complaining about that happening...
The drawback to MS encoding is that LAME is only optimised for stereo listening - if the compressed track is played back through a Dolby Pro Logic decoder, the quality of the rear channel sound can suffer audibly in some cases. In Dolby Stereo, the rear channel is L-R, just like the S channel in MS encoded stereo. LAME only optimises the decoded LR stereo signals for audible artifacts, not the S signal when listened to as is. As far as I know, that is the only scenario where using LAME in LR mode exclusively has been shown to improve sound quality. In all other situations, it performs much better in automatic LR/MS mode, or "joint stereo", so the encoder can decide where to use the bits available.
See this old page for an explanation of MS encoding. There's lots to be found on the topic in Hydrogenaudio's archives, but I can't be arsed to do a search right now.