Slashdot Mirror
Peter Gabriel Wants You to Re-Shock the Monkey
PreacherTom writes "The party line for the music industry has been clear: discourage music downloads at all cost. However, singer Peter Gabriel is taking things in a different direction. In order to promote his own label, he is actually encouraging people to not only download his music, but also adapt it into something more modern. In doing so, he actually posted a sample pack of Shock the Monkey consisting of vocals and other pieces of the original multitrack recording. Some in the music business would call this the commercial equivalent of hiring kidnappers to babysit. In actuality, Gabriel is pleased with the results."
Nine inch Nails put out a track and allowed it to be remixed..
s html?tid=141&tid=3
see
http://apple.slashdot.org/apple/05/04/16/1417205.
They are facinating in how they work, but let me provide a quick laymen explanation:
.wav file with equivalent (for most humans -some one might disagree - i don't) quality.
.j.
First off, your idea that tracks are "seperated" is an understandable mistake! But, the deal is that it's not the tracks that are seperated, it's the component audio frequencies that compose the sound that make up the song that are.
Let's skip the boring stuff and get right to it. If this interests you, i'm sure that wikipedia will have a full explanation. Imagine three people are whistling (and that this makes up the whole, if somewhat boring, song. Person 1 is whistling at 700hz (hertz, or cycles per second. Human hearing is approx 20-20000 hz, rather like the specs you see on headphones, no coincidence). Person 2 is whistling at 703 hz (NOTE this is close to person 1 on purpose) and person 3 is whistling at 900 hz. So you hear, uncompressed three whistles. There are two things that happen to make an mp3:
1) If I can analyze this sound to find it's frequency components for a given "window" (or in mp3 speak, frame) of time, i can just record that. It would be easier (smaller) to say Persons 1, 2, 3 are whistling at 700, 703, and 900 then it would be to record the full sound of them doing it (think about that)
Still, music can be complex, and there are different qualities of MP3 you can make too (usually refered to as bitrate, like 128, 160, 192 Kbps (kilo bits per second) so we have
2) A principal not unlike optical illusions called Psychoacoustics. It basically says that if you have two signals A and B, and A is louder then B, and A and B are close enough in frequency, a person will only tend to hear A. Common sense time, if a headphone speaker is making a sound, and a big loudspeaker is making the same sound, you'll only hear the big loudspeaker. The question is, how much different will the headphone have to be before you hear it?
This is the science of psychoacoustics. Basically, the more compressed an mp3 is, the more will be "stripped" out - that is as the bitrate gets lower, the amount seperating A and B is allowed to increase. On the flip side, if the bitrate is high enough, there is no practical difference to the human ear, because you just can't hear such a small difference anyway That's why a high bitrate mp3 is STILL five times smaller then a
Check on fourier transforms, psychoacoustics, and mp3 on wikipedia for more (and if anyone has a better example, well, typed this pretty quick, go for it!)
CS majors know the time/space tradeoff, but they never get taught the 3rd, crucial, tradeoff of the set: comprehension!
Others have provided useful aspects of the answer to your question but I don't think anyone has boiled it down yet.
In short - No. A single track compressed will work better in mp3 than individual tracks mixed together.
The reason is that mp3 is designed precisely to compress single multi-instrument tracks and makes use of psychoacoustics to do this. The gist of which is, the more complicated a sound is (multiple instruments/frequencies) the less of each individual instrument (frequency) you are likely to be able to perceive. Thus, with all the instruments together in the one track, the mp3 algorithm can work better to strip out the subtler elements you don't perceive. If you are just compressing a single instrument there is less of that compression that can be done because, for example, it doesn't know that the rhythm guitar is being drowned by the kick drum at that point in time. Or as a corollary, compressing a single instrument will have to remove stuff you can hear just to hit the same bitrate as the compressed single track. So, combining individual tracks will lead to a worse outcome, all other things being equal, than compressing the already mixed track.
Sure, the RIAA could dig some plain-old selling-CDs value out of it, but they've gone to that well plenty of times. So this is as much publicity stunt as artistic endeavor, and it's reaffirming exactly what the RIAA does: promote big acts.
Peter Gabriel is British. He has converted a garden shed on his own property into a recording studio where he produces for his own label. He actually runs his own website.
Yes, he's a big act, but since leaving Genesis he's been as much as possible an independent big act publicly at the forefront of not paying too much mind to copyright issues.
When his "people" came to him all upset that people in India were pirating his records his response was (paraphrasing):
"You idiots, book me. If they're not paying for what we're trying to sell they're at least demonstrating a demand for what we can sell that they can't pirate."
He has a long, personal history of being the good anti-Metallica.
KFG
Actually the Fairlight was 8-bit, as were the Emulator, Emulator II and Ensoniq Mirage (among others). The latter is quite interesting in that the DOC chip (which handled the sample playback) found its way into other Ensoniq synths (where short waveform loops were held in ROM) as well as the Apple IIgs. In the Mirage and Ensoniq ESQ and SQ range it fed into rather nice 24dB/octave analogue filters, giving a very rich and warm sound from the fairly low-quality samples. The PPG Wave series of wavetable synths also used 8-bit samples, again fed through a 24dB/octave filter. It was only a little later that 12-bit sampling became common, with Akai, Casio, Emu, Sequential Circuits and Ensoniq brought out 12-bit samplers (again, Ensoniq at least also used the voice chips in wavetable-based synths), then ultimately on to 16 bits.
Oh, and let's not forget the many 8-bit parallel port sound devices in the early 90s like the Covox Speech Thing, Disney Sound Source and others.
You are all complete idiots and do not know how MP3 compression works. I cannot belive this trash got modded up.
You in particular have just mixed up HZ and KHZ and injected more bullshit like "It essentially tries to fit a curve to the master waveform".
Perceptual encoding is much more complicated than that.
It actually performs an FFT analysis and split the sound up into it's component sine waves.
Then, two methods are used to discard data.
Both known as perceptual masking. The first method deals with frequency masking, the second with time.
Human auditory perception cannot hear a quiet frequency when there is a louder one within a few hz of it.
So, you can discard all of them.
Humans cannot hear a quiet sound when a louder one immediately follows it. (Think of a bass drum, you do not hear the squeak of the pedal just before the beater hits.)
So you can discard all those too.
The watery effect of heavy MP3 compression is from too many transients being discarded by the second method, so the transients appear spread over time. The thin lack of depth is due to too many frequencies being discarded.
"the net result of a sound around 701.5 Hz coming in and out every 1/3rd of a second. It would basically sound like 3 beeps a second, though more like a siren than a beep. If the waves were at different amplitudes, the same phenomenon would still exist but there would not be complete silence during the destructive phases."
This is crap. The cancellation has ALREADY HAPPENED when the waveforms were mixed before you do the MP3 compression. So you just need to compress the result, not the individual tones.
Also, it will sound like an amplitude tremelo, not a siren which would imply pitch modulation.