Got a song I've put a lot of time into and now that I'm working on the vocals I'm realizing it's a bit draggy and needs to be sped up just a tad.  It's not a difficult song to play or anything, and I usually perform it faster anyway, but it's lot of guitars and the sound is right where I want it.  I tried using the Cakewalk Audio "stretcher" (in Home Studio 2...yeah, I know, really old version) and it keeps the pitch ok but there are all kinds of artifacts and skipping and is just completely unusable.
 
Is there anything any of you could recommend that does a transparent job of speeding up the audio just a bit, without changing the pitch?  I just don't have the energy and the time to re-record all of those tracks just to bump the tempo a bit.  
 

When you do any type of stretching like that it will use the "online" algorithms which are meant only to give a preview of what the time change/pitch change/etc will sound like. Once you "bounce" the stretched clips(s) it will use the higher quality "offline" algorithms which will clear up all those artifacts you are hearing.
 
Here's the thing though. There are multiple "offline" algorithms to choose from and some work better than others on specific types of material. Generally the RadiusMix once work well. IIRC you can set these in the Clip (or Groove Clip) Inspectors (which I'm now realizing you may not have access to in such an old version of Sonar).
 
Anyway... I have no idea what Sonar 2 has as far as features for this stuff but definitely you're gonna want to bounce the stretched clip so it gets rendered using whatever "offline" algorithms you have available to you.
 
There are also many other methods to stretch and fiddle with stuff that can be tweaked like the Loop Construction View (if you have access to that in your version) where if there are artifacts even after bouncing the stretched clip (Acidized Groove Clip) you can adjust the transient splits and whatnot to make the stretch sound better.
 
Hopefully you've got a manual around to see how this is done in Sonar 2 because I only know X series stuff.
 
Cheers.

Thanks Beepster.  I went back and bounced the guitar tracks to one stereo track.  Then I stretched the audio by 95% and that seems to be about where the tempo feels right now.  Artifacts were present and wave view looked all weird, as per usual, so I did what you said and bounced that track to yet another one and sure enough it cleared up the wave view back to normal, saw the track was a little shorter and the artifacts are mostly gone but it still sounds "weird".  
 
Something still isn't right.  Feels like some notes aren't *on* here or there, and I can still kind of hear some saturated syncopated effect on open chords and such.
 
I wonder if my Cakewalk is too old for any quality audio stretching to be done, en mass with such a thick guitar sequence...?    My version is so old they don't even call it Sonar.  This was sold back in 2001 or so, I think.  I can't upgrade to Sonar anything without upgrading my whole PC.  This is a Windows XP machine running this old version and it just works.  I'm in no financial condition to do any of that, and won't be for quite some time.  Bummer.
 

 
 
 
 

Edit: Before attempting ANY of this make sure you do a "Save As" of your project so you don't lose your original project as you play around. Also I usually make copies of any clips or tracks I intend to do things like stretching on so the originals remain available so I can try again without using "undo" and or I can try out different techniques and compare the results (and choose the best ones). You probably know this but I'd hate to lead you to a crash or have your raw waves get screwed up because I urged you to get all fancy arse.
 
Original Post...
=======================================================
 
Oh... yeah, you should not be bouncing all the tracks together before doing the stretch (ideally). Stretch your individual mono tracks individually (you could stretch them all at once if your computer can handle it but either way... leave them as separate tracks when stretching). Then "bounce/render" the stretch to each individual track on at a time (still separately).
 
That should work better. A lot of stretching stuff relies on the transients in an audio clip so by bouncing/mixing all your clips together you likely made things harder for the stretching stuff to really detect the transients for the stretching (like the multiple clips created less precise "splits" at the transients so it futzed up).
 
As automatic as this stuff is these days it is still a rather precise procedure that requires your splits be at the best spots on the clip (at the transients) to work well. So if you do them separately it should work better BUT if not see if there is something called a "Loop Constructor" view which allows you to get right into the wave and adjust where the splits occur (like sometimes the transient detection will miss some transients in which case you add them or it detects the transients too early or late in which case you move the splits to where the actual tranients occur).
 
That is more "GrooveClip Looping" stuff (as opposed to the Izotope stretching done in Track View) but it can acheive the same thing.
 
Again it depends on what features that old version has but if you got this far I can only assume that you have the tools you need to make it happen and doing whatever stretching you did here on the INDIVIDUAL tracks and then bouncing the INDIVIDUAL tracks one at a time you'll get less artifacts.
 
Look into the different "offline algorithms" you have access to as well.
 
Sorry to be vague but as I said... I'm an X series brat so I can only guess as to what you have access to. Check the manual for some of the terms and techniques I've mentioned here.
 
Also know that the more drastic the tempo change the more problematic it becomes. So if you are doing something nutty like going from 100bpm to 150bpm or 200bpm the results will be less desirable and/or you will REALLY need to get in and place your splits accurately by hand in the Loop Constructor (if you have it).
 
If you are only increasing/decreasing the tempo by say 5bpm's then it's usually going to work out better/easier.
 
Also disable ANY and all effects before bouncing (or disable them from the bounce if possible) because transient detection works best on clean, dry, articulated tracks and let's the software work better. You can add the effects back in after you have bounced.
 
Certain dynamic effects though may actually HELP define the transients better such as compressors or limiters that even out the peaks a little but if you totally squash it (so the lower parts are closer to the peaks and make the transients less defined) then it starts to not work so good.
 
Basically heavy distortion or tons of reverb is back for transient detection. Something like a limiter that accents the transients a little better MAY help... but due to the bounce you may be stuck with it.
 
Ideally you want a nice, clean and cleanly played part with defined transients (in the case of guitar cleanly picked notes and chords).
 
Hopefully this points you in the right direction. It's a bit of a confusing topic but I have been VERY impressed at some of the results I've gotten with Sonar using some of these principles.
 
Cheers.

Thanks again, Beepster - that makes total sense, I'm trying it now...  
 
No worries my friend, I already did a 'save as', in addition to the duplicates of the originals I always keep in case one file crashes and becomes unrecoverable (I had a song once that became inaccessible after an audio engine error during playback).  
 
 

You're going to experience artifacts stretching audio by 95% regardless of what software you use to do it. That's almost equivalent to lowering your sample rate from 44.1 KHz to 22.05 KHz and then interpolating all the lost data.

Beepster - I tried it like you advised, each track stretched separately, then bounced separately, and there's still all kinds of issues.  I've resolved to just re-record the damn thing.  Really bums me out, but I can't have that draggy tempo and I can't suffer the artifacts from this process so...
 
bitflipper - I found the percentage to be a little confusing.  95% (5% shorter than the original) only makes the whole 5 minute piece about 15 seconds shorter, overall.  If it's that impacting on the data, then yeah, there's no getting around the artifacts.  
 
I guess now I'm wondering...if my stretch is that mild, and it sounds like that, then what use is stretching at all?  Why is there a scale of 25% to 400% if even just a 5% deviation from original is that terribly rendered?  
 
At any rate, thanks for all your help Beepster.  I'll just have to redo it.  
 

bitflipper
You're going to experience artifacts stretching audio by 95% regardless of what software you use to do it. That's almost equivalent to lowering your sample rate from 44.1 KHz to 22.05 KHz and then interpolating all the lost data.

If the tool OP is working with is similar to the clip Time Stretching in newer versions that 95% might actually be a 5% speed up. Like when I use that method (the one where you hold control and slip edit the clip) it starts at 100%. If you speed up the clip slightly (by making it smaller) it'll use 100% as the base and gradually go down by percentage points. 95% in that scenario only represents a slight speed up. 105% a slight slow down.
 
That is of course the Sonar 2 stretching operates in similar manner. It may not. An actual 95% speed up (so 100bpm to 195bpm) would definitely be problematic. I've gotten away with some pretty drastic changes though without artifacts but it still had an "odd" quality to it when using the slip stretch.
 
However, and I have yet to try this to such extremes, in one of the Anderton X1 Advanced video series he manages to get those drastic time stretches (like doubling/halving the speed or more) using the Loop Constructor. In fact he recently typed up that technique for his Tips thread in the Sonar forum (it's part of the Varispeed technique... if you ignore the pitch aspect of the trick). It's all about getting WAY into the transient slices and requires the audio to turned into GrooveClips.
 
I have done a bit of tinkering with that when I was playing around with some of the sampler VSTs and it does seem to work very well without aritifacts but it's a LOT of precise transient futzery.
 
I find it all really fascinating and this type of stuff is going to be very useful to me for fixing up sh*tty live off the floor demos recorded without clicks.
 
Cheers.

Voda La Void
Beepster - I tried it like you advised, each track stretched separately, then bounced separately, and there's still all kinds of issues.  I've resolved to just re-record the damn thing.  Really bums me out, but I can't have that draggy tempo and I can't suffer the artifacts from this process so...
 
bitflipper - I found the percentage to be a little confusing.  95% (5% shorter than the original) only makes the whole 5 minute piece about 15 seconds shorter, overall.  If it's that impacting on the data, then yeah, there's no getting around the artifacts.  
 
I guess now I'm wondering...if my stretch is that mild, and it sounds like that, then what use is stretching at all?  Why is there a scale of 25% to 400% if even just a 5% deviation from original is that terribly rendered?  
 
At any rate, thanks for all your help Beepster.  I'll just have to redo it.  
 

That is disappointing and really should not be happening for such a slight change. Perhaps it is simply because it's so old and the algorhithms can't handle the material.
 
It's quite possible that back then it was only intended for use on more percussive material like drums which respond to stretching much better than things like guitar.
 
At some point Cakewalk introduced iZotope time warping algorithms for this stuff (if you are familiar with iZotope you know they make top notch stuff that almost seems impossible). I'm guessing that if those aren't being used in that old stretch tool that would explain the difference in your experience and mine.
 
Alternatively maybe the process of doing the Offline rendering is different in Sonar 2 so the rendering of the stretching isn't happening (or isn't happening in the best possible way).
 
Actually make sure you don't have your Offline render/bounce settings to be the same as your Online render settings. In newer versions it allows you to render using the crappier, artifact filled Online algos but there are a half dozen WAY better ones (like RadiusMix and the like).
 
Retracking can sometimes suck but look at it this way... the parts will probably come out a lot better now that the tune has settled into your mind. I know if I retrack something it pretty much ALWAYS turns out much better.
 
And not to sound like a shill but if you can swing it you might want to consider upgrading. The new Sonar is really freaking snazzy... but you'd have a HUGE learning curve to catch up.
 
Good luck.

Beepster, you are confusing SONAR 2 with SONAR Home Studio 2. Both are older products, SONAR 2 was released around 2002 and SONAR HS2 around 2004.
 
This is a free plug-in which might still work given the age of your DAW. IIRC, it was fairly popular at the time. It is called Delaydots Pitchworks. It can do both pitch and tempo shifting.