Ableton Forum

Angstrom wrote:Even though Elastique Pro is better than 'efficient', it is still not artifact free, it is just a better featured real-time stretch than 'complex'. It is not a quality offline stretch.

Ah, I was wondering about that - i.e. whether Elastique Pro would tick both boxes. Shame it doesn't look that way...

In my opinion, a better offline algorithm is even more important than a better real-time algorithm. I can get away with the issues with the current algorithms during a live performance, but I can't use them for serious production.

yep,
I bet a lot of people thought that 'Pro' was equivalent to an offline stretch. Hence my little joke about the popularity of this thread being mainly because it has "Pro" in the title

I only posted this wish because Live already has the base level technology from Elastique so they may as well move up a level as processors have improved in the intervening time.

An offline stretch is a good idea too, a different good idea.
It would preclude tempo-shifting while that 'offline' mode was selected because each 'offline stretch' clip would need to generate a hi-quality wave in the project folder.

It would be a 'studio' stretch of course, but as the majority of the users do use Live in a studio setting it would be worth adding it. While "Elastique Pro" is more for the live situation.

Nevertheless, the distinction between a "realtime" and "offline" timestretch isn't simple, as the stretch algorithms usually thought of as being "realtime" aren't (of course) truly realtime in the strict sense, otherwise they would be able to see into the future when speeding up material . However, they are functionally realtime enough to allow smooth parameter changes. Similarly, the "offline" timestretches often complete their task in considerably less time than the length of the material they're stretching, and to some extent they can be integrated into a sequencer/multitracker to appear to function in "realtime" on a given clip, too.

The point is, if a timestretch algorithm is implemented in an offline manner (throwing a progress dialog and making you wait for the transformation to be over), it isn't by all means a guarantee that it would produce a better result than a perceived "realtime" one.

That being said, I'd be very interested to know what offline timestretchers are there nowadays that use considerably heavier calculations than the algos in "realtime" applications? How much better do they sound, and are there demos available?

I think the primary distinction between what we call 'realtime' and 'offline' is : if you waggle the tempo about will it say "RECALCULATING" or will it continue to play?

All of the Elastique algorithms continue to play, so in my book they are 'realtime'.

I don't know what the current state of offline processing is in this area. Any stretch will give artifacts, I mean you can't make something out of nothing without making some data up!
But I don't know if there is some new magic 'beard-based' voodoo that is especially transparent.

Angstrom wrote:I think the primary distinction between what we call 'realtime' and 'offline' is : if you waggle the tempo about will it say "RECALCULATING" or will it continue to play?

Yep, I agree, the functional end result (the editability of parameters during playback) is what counts. To be sure, I edited my post to be more specific while you were answering.

The quality difference is what I'm getting at: if a stretch algo is implemented to work completely offline, it might still use less computational resources for the actual stretching than a realtime-functional algo, or sound otherwise inferior. That's why I'm interested to know whether there are some truly superior offline stretchers around these days, like, ones that would take their time and crunch on a track for a long time, producing stellar results, utilizing the kinds of beard-based techniques you were referring to .

I would love a higher quality time stretching algorithm, even at the expense of requiring greater processing power.

It's good for elastic bands.

+1 Pleeeez

+1

Especially with CPU's getting faster and cheaper, it makes sense.

+1

+1
come on fellas

+1

+1

+1

-1

leisuremuffin wrote:please no.

I already *HATE* complex mode. I don't want anything else from that company in my software. It's a cpu hog, it sounds like shit, and in my experience it's been the number one culprit in causing "skips" in audio on my setup. I NEVER use complex mode anymore because of this.


Whatever, if you guys all want it so bad, fine. But I really don't and don't want to have to pay for it, and i don't want the abes to support that company. In my opinion it is a shit product.



just another opinion, that's all.

from the thread in general section. Thought i better represent here as well since the abes are more likely to read this.




.lm.