Justin Bieber 800% slower = Ambient Masterpiece
- Thread starter zen5887
- Start date
Recommended Videos
Oh wow that's pretty awesome. Makes me wonder what other songs could benefit from this... I'll get back to you on that.
Thank youTimeLord said:http://cdnfiles.info/static/USMILEAMBIENT.mp3aNimeKing33 said:
Right click and save as
Conspiracy theory - a theory that explains an event as being the result of a plot by a covert group or organization; a belief that a particular unexplained event was caused by such a group.CaseySmith said:
The event of this song sounding good when slowed down 800% is the result of a plot by JB to covertly hide great compositions in his songs. I believe that this unexplained event is the result of JB's secret composer side.
=P
If you amassed a little effort, you would have simply looked at the comments on the page where there's a mediafire link.lacktheknack said:
But it isn't. Speed it up and it ain't the song in the original form - it's heavily modified with DSP. So it wasn't a matter of just "slow it down, it sounds like this". Slow the song down 8x with a proper time-stretcher that doesn't add digital artifacts or chop out large chunks of the signal in order to "smooth" it over, and it'll sound nothing like the result in the download.Mad Stalin said:
Damn. Well, I shall repeat myself once more then!Kragg said:
Holy. Fucking. Shit.
I have been mind-fucked, haven't I? Where is the mind control device, I demand to know!
I was just thinking out loud about how I might approach emulating that effect, without using specific standalone apps. The answer to the question I put forward, was, as it turns out, to use the standalone application: Paulstretch.James Joseph Emerald said:
I don't think you're quite right about reverb being a natural effect of applying time stretches.
Simply taking a song and slowing it down, as much as the 8:1 ratio of the JB example, would just turn it into low-end DC offset sludge, because in terms of pitch (without correction/preservation) algorithms, playing the song that many times slower would be exactly equivalent to tuning it 8 octaves down a keyboard. This is called re-sampling... it might sound weird, but it doesn't actually generate artifacts (or sound particularly epic, by itself).
As far as pitch preservation is concerned, generally it's approached one of two ways.
The first, is "granular" where the audio clip is mathematically divided into extremely small 0.x-millisecond "grains" which are copy pasted in sequential order, algorithmically, to artificially lengthen a sound sample without altering it's pitch. The result does not sound like reverb, but rather like the buzzing digital sound effects you hear a lot in "The Matrix", for example. There are definately audible artifacts the more extensively this process is used, but it sounds more "fax machine" than "worbly" and not at all like the JB example.
The second, is a much more CPU intensive algorithm implementing spectral analysis and resynthesis, usually based of mathematical models developed by Joseph Fourier - hence FFT or "Fast Fourier Transforms" (most commonly used). This analyzes the amplitude, or "loudness" of (128, 256, 512, 1024, etc) "bands" representing an increasingly specific resolution of harmonic partials (sine waves) over time. The analyzed results can then be applied to govern the amplitude of a new, repitched, lengthened, or shortened (theoretically infinite) number of new sine waves (each with an individual value of Hz, between 20-24,000 [human range of hearing]) to create a newly resynthesized version of the original sound with new time and pitch specifications. This is probably the the most "worbly" and sometimes "reverberated" sounding form of pitch shifting, but it has a certain characteristically "synthetic" sound to it's processing that, again, isn't present in the JB sample put forward.
Ultimately, as I said at the beginning of the post, it's the GNU public liscence Windows and Linux application "Paul's Extreme Time Stretch" or "Paulstretch", that was used on the JB sample. While I can't explain exactly what's going on under the hood, mathematically, I believe, in a general sense it would be true to describe the main function of the application as a "[Multi-Buffer-Window]-as-crossfading-'Macrogranulizer'", in terms of how it works.
P.S. - Basically what Bonsaik said, Madstalin.
Being interested in audio engineering and recording software I can tell(as has been mentioned) this is certainly not just a song 8X slower than usual. Nor do I particularly enjoy what it is.
