Good question, actually. The way I see it is this. The most important element in getting AudioSnap to work perfectly is to have a perfect transient map. Whether AS can do this or not depends on the track itself. At one extreme, if it's a four-on-the-floor kick drum, AS will create a perfect transient map. At the other extreme, if it's program material the transient map will be pretty much useless.
 
Now, I freely admit I also have a philosophical prejudice, which is not to correct something unless it SOUNDS wrong. Therefore I listen to a track. If something sounds wrong, I fix it manually by either grabbing an existing transient and moving it manually, or inserting a transient and moving it manually. No beat/transient corrector can decide what sounds wrong, only what it thinks looks wrong. 
 
However, I have used AS in "auto quantize" mode. For example, I was doing a remix of a dance tune that had a recorded electronic kick drum. I wanted to layer a drum loop played by a human, but when the two kicks hit, there was often flamming. I copied the human drum part, filtered out the highs, used the transient shaper to give real strong transients, and applied AS to create a pool. The transient map was perfect, and it was painless to then snap the electronic kick to the pool and sync up with the human drums.
 
There was also a rhythm guitar part where I wanted to have metronomic precision because it was being played against electronic drums. So I "auto-corrected" the part with AS. When I played it back, there were two places where AS really screwed up but the rest was fine. So I hit "undo," split the clips with the problems, applied AS to the rest, then fixed the two problem areas manually.
 
So the bottom line is this. AS can be a "magic" solution, but most of the time you'll need to at least edit the transient map. How much you'll need to edit it depends on the track. Rather than take the time to create a perfect transient map and auto-correct, I'll just use the Transients edit filter, fix what needs to be fixed, and be done with it. This can take anywhere from a few seconds to several minutes per track. 
 
Finally, it took me a while to develop AS chops. I didn't use it for a long time because it was just too much hassle and I never got really good results. However, in the course of using it, I've learned what it does well, how to compensate for what it doesn't do well, and developed my AS skills to where I use it frequently and (I think) effectively.
 
Does that help?
 
 
 
 

For AS beginners (I count me in that group) try extracting midi hits for the kick drum track and sending it to a drum synth to mix with the real kick. That's pretty easy to do.

I use it. I think it's very good. BUT I think it doesn't do much well automatically. You have to go in and do a lot of manual work to get good results. For example, forget about using the quantise function unless you're willing to zoom in and fine tune the transient positions. That's got better in the latest release, but it's still not perfect.
 
Audio Snap's real weakness is in workflow design. It's full of obscure gotchas and opaque procedures.
 
 

John T
[...] BUT I think it doesn't do much well automatically. [...]

I think that sums up AS in a nutshell. AS is a big toolbox that requires user know-how to get the best results from. Very little "point and shoot" functionality, but for those who have taken the time to learn it well, they can do quite a bit with it.

My main use for audiosnap is extracting tempos from freely played material. Most of my songs start life on acoustic guitar or sometimes piano and then I use that as the framework for the song to build on and then disregard the scratch track once I've got the timing and feel out of it.
 
I've done much the same when recording bands but usually take the timing from the kick and/or snare and go from there.
 
I'd also +1 on the fact that it's not a simple one click tool but a bit of time understanding how it works and the tools available was time well spent for me.

Anderton
 
...For example, I was doing a remix of a dance tune that had a recorded electronic kick drum. I wanted to layer a drum loop played by a human, but when the two kicks hit, there was often flamming. I copied the human drum part, filtered out the highs, used the transient shaper to give real strong transients, and applied AS to create a pool. The transient map was perfect, and it was painless to then snap the electronic kick to the pool and sync up with the human drums.
 
 

I don't use AS all that often, but I really need to remember that technique!   It's a good one.
 

You need a degree in AS to master it .. Personally I haven't used it in so long , it's just like ACT is a waste of time and workflow killer . Add this transient.delete that one , use it like this or that , seriously !

djjhart@aol.com
You need a degree in AS to master it .. Personally I haven't used it in so long , it's just like ACT is a waste of time and workflow killer .

 
Just like AS, ACT can be used in a more basic manner. This blog post describes how. The key is to take the time to do the initial setup correctly. After it's set up, you can use ACT for easy access to whatever you want to control.
 

Add this transient.delete that one , use it like this or that , seriously !

It's important to remember that beat detection processes started with Digidesign's Beat Detective, which was designed specifically for dealing with material that had distinct transients, i.e., drum tracks (and preferably multitrack drum tracks). AS is no different: that's where it works best. The more the program material deviates from something with a strong beat and obvious transients, the harder it will be for an automatic process to decide what is "signal" and what is "noise."
 
The bottom line is if the timing needs fixing, the tradeoff is how long it would take to (a) re-record the part to get the timing right vs. (b) how long it would take to create a transient map so AS can do the correction for you vs. (c) how long it would take to listen to the track and use AS in conjunction with the Audio Transients edit filter to correct those elements that sound wrong. For me, (c) wins 95% of the time.