It's a product showcase for Melodyne 2, a program that apparently will do the "impossible." They have a demonstration of a Mozart string quartet, being broken apart into individual notes and convincingly altered before everyone's eyes. The processing power and programming language to do this kind of thing in a workable amount of space and time has arrived. I;'m absolutely floored by it. It's going to completely change how a lot of producers make popular music. Classical composers will do even more frightening and awesome things.
For one thing it's going to once and for all sound the death knell of the outdated Midi protocol. It may even completely overfly Open Sound Control to simply allow musicians to access sampled notes and manipulate their qualities in real time. This is essentially the launch of Sputnik to me: someone has finally done it.
Keep in mind, when it comes to a music programming demonstration, what you see is 1% of what you'll eventually get. Often times the programmers have no idea how the programs will be used in the end.
Posts: 9912 | Registered: Nov 2005
| IP: Logged |
posted
That's amazing. You'd never be able to tell that it was digitally modified! It's so bloody clean!
Posts: 3295 | Registered: Jun 2004
| IP: Logged |
posted
If I had the money, I'd buy a mac for this. That is such a sweet toy!
Amazing application for music classrooms, demonstrating the nuances of different scale modes by converting music on the fly in front of your students, wow.
Posts: 3846 | Registered: Apr 2004
| IP: Logged |
quote:Originally posted by TomDavidson: You'd have to get a Mac.
In this specific case yes, but the plug-in is what they're really interested in distributing, and it's only a matter of a very short time before this technology is licensed across the board, or just stolen.
Just as a word to the wise, it's not going to sound perfect right away, so we will probably be subjected to the usual treatment of producers using the technology in situations where it really isn't warranted- the creator acknowledged that as of now, only certain kinds of sound work well. Still, this is a tremendous step.
Posts: 9912 | Registered: Nov 2005
| IP: Logged |
posted
MC, I'd go farther than that- and I wouldn't say "songs," but I'm a snob.
We essentially have the equivalents of photoshop for music already available to us, we've had it for years, with the ability to edit many of the qualities of the sound en mass. This tool is going to allow people a creative power to focus on the qualities in music that are inherently different from images- because up until now photoshop is essentially how many sound editors have worked, much of the toolkit was the same. I've done a lot of the stuff that this guy was demonstrating on the screen, but I've never done it in 10 seconds, and I've never done it with the complexity of source material he presented.
The interesting thing here is not going to be how this is utilized for looping and song writing, because that element just imitates a process that could be done more slowly before, but how this program allows us to analyze sound environments and really, really get into the nuts and bolts of a particular group of sounds and learn how it works. As someone who spends hours a day in programming environments trying to figure out what makes a particular sound tick, I am excited.
I'm keeping my eye on this and Jitter, the cousin program to Max/Msp, which handles video. I would not be surprised if we start seeing CGI type animation start to be outdated in the next ten years with powerful video sampling and arrangement techniques- essentially the transformation from MIDI (not real sounds, or only a limited set of sound elements) to this process, utilizing natural input to create real new environments. Think of a program that could analyze video of an environment, and quickly create that environment as a template for transformation- you wouldn't need to code all that stuff, you'd need a program fast enough to do it for you.
Posts: 9912 | Registered: Nov 2005
| IP: Logged |
posted
Yeah... more than photoshop, it's more like what the Wachowskis did in Matrix:Reloaded, except in real time.
The part where he took the little guitar lick with the chuckas and harmonics and then transposed it in real time via keyboard was very impressive... and I'm willing to bet you that two thirds of what he then played was impossible to play on a guitar, but it sounded natural as can be, as if some guy was standing there changing his fingerings, like Neubaker said.
It's not like photoshop, it's like loading digital video into your computer, having the computer analyze it for a few seconds, and then being able to click and drag individual objects around in real time as the video plays.
Posts: 3846 | Registered: Apr 2004
| IP: Logged |
I think this also bodes very interestingly for the future of digital video- the development of these technologies usually if not always favors music at first, even if the music part of it is always lower-profile than the video side. The video applications this stuff will effect will reach a wider audience, but music will get there first.
AS for the guitar part, yes playing those notes in real time would be a next to impossible task, especially when natural harmonics are used.
Posts: 9912 | Registered: Nov 2005
| IP: Logged |
posted
Wow... Just the other week I was thinking "wouldn't it be cool if computers could do this?" and then this thread came along...
Posts: 1029 | Registered: Apr 2007
| IP: Logged |
quote:For one thing it's going to once and for all sound the death knell of the outdated Midi protocol.
Well, although I think this is a great advance, I've heard that before with other technologies. It also reminds me of similar statements about the future of vinyl.
Very cool, however. Very cool.
Posts: 496 | Registered: Apr 2005
| IP: Logged |
posted
Achilles, yes that's true- midi is like fossil fuel, it works, it's everywhere, it's still awful, and we're straining for a change.
Posts: 9912 | Registered: Nov 2005
| IP: Logged |
posted
I saw this a while back. It's impressive, but they only demo very clean sources. I'll be curious to see if it will also work for separating the voices in a metal band, for example.
If it's as impressive as it seems I may very well buy the Logic plugin when it comes out, since I have Logic Studio at home.
Posts: 10886 | Registered: Feb 2000
| IP: Logged |
posted
Almost you have persuaded me to be a Mac user.
When I get a new computer in a few months, if this software exists on PC I'll pay whatever it takes to get it.
Posts: 14316 | Registered: Jul 2005
| IP: Logged |
It would seem a logical extension to me that this software could then be used to transcribe a passage of music into engraving software of some sort like Finale or Sibelius. The quantizing of the rhythms would be an issue, and I'm sure it would be necessary to add in markings like slurs, accents, etc., but it would be a useful bit of software.
Posts: 1099 | Registered: Apr 2005
| IP: Logged |
quote:Originally posted by twinky: I saw this a while back. It's impressive, but they only demo very clean sources. I'll be curious to see if it will also work for separating the voices in a metal band, for example.
Probably not. The cousin tool to Melodyne is called "fiddle" and it was designed in Max/Msp, and unlike Melodyne it is free, once you have MSP. My teacher works with fiddle, which is probably what this plug-in is built on, and it's very very good, but doesn't do distortion well at all. And really, who can blame it? If you were just sitting looking at a frequency spectrum for a metal clip, you'd have a very hard time figuring things out- these analytical programs key in on simple information to build a case for one note over the other, but distortion basically ups the ante by a factor of ten in that task.
Posts: 9912 | Registered: Nov 2005
| IP: Logged |
quote:Originally posted by BlackBlade: Almost you have persuaded me to be a Mac user.
When I get a new computer in a few months, if this software exists on PC I'll pay whatever it takes to get it.
I would look into a program called Max/Msp from Cycling 74, in San Francisco. It's a small company, but the program is the most respected, most used programming environment for musicians, and it's available on all platforms. The tools used to create this program are available in different, less evolved senses, in Max/Msp, and you can build your own tools with max that do, essentially, anything with music you could think of.
the problem most people run into is that these music programming environments, including melodyne, are a blackhole of features. There is so much to do that you will never, ever, learn everything, so you adapt the programs and tools to your own repertoire, rather like an instrumentalist develops a playing style.
Posts: 9912 | Registered: Nov 2005
| IP: Logged |
quote:Originally posted by BandoCommando: Not bad.
It would seem a logical extension to me that this software could then be used to transcribe a passage of music into engraving software of some sort like Finale or Sibelius. The quantizing of the rhythms would be an issue, and I'm sure it would be necessary to add in markings like slurs, accents, etc., but it would be a useful bit of software.
There are tools available already for this, but there are many many bugs and barriers against a useful comprehensive application. Basically, the technology all exists, but working out the bugs, all the possibilities for something to go wrong, all the automation that needs to be set in motion, it's a daunting task. In a few years there will programing AI that will help to overcome the sheer volume of work to be done in these fields, but that isn't around just yet.
Still though, I think you're going to see that really scary and creative AI is going to be developed in the arts before something like "skynet" appears. The arts are where creative solutions are needed, defense is simpler.
Posts: 9912 | Registered: Nov 2005
| IP: Logged |
posted
I would be really, really surprised if your iMac could run that program with any kind of speed.
Posts: 37449 | Registered: May 1999
| IP: Logged |
posted
I'd be surprised if it couldn't. It looks like current versions of the software run on Macs 1 Ghz or above, PPC or Intel. Current iMacs can be had at around three times that speed (or more, if we're counting from the PPC benchmark). The capability will be available in all of their software, so unless they're going far, far beyond the existing system requirements, it'll run just fine.
It might take longer to run the initial note breakup. We saw it take a few seconds on their demo system, but I doubt it will take more than a minute to perform the initial analysis of the sample on a modern iMac, and that's assuming a huge amount of multiple processor optimization. Of course, that part of the process is just a small slice of the time the user will work with the capability, so that's basically irrelevant (to everyone who really wants the capability).
Posts: 15770 | Registered: Dec 2001
| IP: Logged |
posted
A slower machine might have serious trouble with the real time pitch manipulation via keyboard...
Posts: 3846 | Registered: Apr 2004
| IP: Logged |
posted
Could be. I'm not sure, though; it looks like they build an internal model of the music during the part that takes longer, and that afterwards they're just manipulating that model in fairly straightforward ways.
Also, I'm not sure how they'd take advantage of multiple processors for that portion, due to the indeterminacies multithreading introduces. I bet it only takes one core, and in that case, an iMac's cores don't have to be much slower than a Mac Pro's cores.
A machine with truly slower clock speeds might start having difficulties. I suspect they'll work on it a lot so the low end iMac and high end Macbook Pros can all handle it (more the latter than the former, but the former comes along with the latter). A lot of their target audience will have or want to get Macbook Pros.
Posts: 15770 | Registered: Dec 2001
| IP: Logged |
posted
I don't know what the system requirements might be, but I have a pretty decent computer. Intel 2.4 ghz core duo, 2 gigs ram, 256mb ati video card, 320 gig HD.
Not exactly the best out there, but definitely not bottom rung Tom.
Posts: 4229 | Registered: Dec 2002
| IP: Logged |
Printer-friendly view of this topic
![[Wink]](wink.gif)
![[Smile]](smile.gif)
That is one of the coolest things I've ever seen a program do.
