Increased sample rates + time stretching (Elastic Audio)

[Morgan C]

I use Sonar's AudioSnap for my drum editing, which involves stretching all the audio between transients. I find it works better than cutting and crossfading, and uses less CPU too.

So I was thinking, because there will obviously be some loss of quality during the stretching, will recording at higher sample rates and then bouncing down to 44k after editing help at all?
Because higher sample rates will mean more samples per second, so when stretching it won't be stretching samples to fit, but just moving them around and then deleting the unneeded ones when I bounce it (badly explained)

 

[PhilR]

It would depend on the algorithm I'd expect. I've had to experiment with the Elastique setting in Reaper to find the best sound for drum editing.

 

[Erkan]

It would depend on the algorithm I'd expect. I've had to experiment with the Elastique setting in Reaper to find the best sound for drum editing.

Is it even possible to edit drum tracks by audio warping in Reaper?

I always thought that wasn't possible.. if it is, please explain!

 

[PhilR]

Is it even possible to edit drum tracks by audio warping in Reaper?

I always thought that wasn't possible.. if it is, please explain!

It's a couple of clicks longer-winded but still pretty easy. Group your drum tracks, split at the transient you wish to "warp". Hold down the "Alt" key to time stretch then simply drag the segment edges to the grid. Make sure you have "auto fades" set to zero and the edges of the waves will still line up sample-accurate without needing to crossfade. This is basically identical to how audio warp in Cubase works, Cubase merely does away with the need to split the track. Plus in Reaper you can use the technique on as many tracks as you want.

I've been working on my own bands drum tracks this way. It's not fast by any means but I prefer to do it this way.

 

[Machinated]

It's a couple of clicks longer-winded but still pretty easy. Group your drum tracks, split at the transient you wish to "warp". Hold down the "Alt" key to time stretch then simply drag the segment edges to the grid. Make sure you have "auto fades" set to zero and the edges of the waves will still line up sample-accurate without needing to crossfade. This is basically identical to how audio warp in Cubase works, Cubase merely does away with the need to split the track. Plus in Reaper you can use the technique on as many tracks as you want.

I've been working on my own bands drum tracks this way. It's not fast by any means but I prefer to do it this way.

of course you are entitled to your own ways, but as something you may want to try:

do the same method, but instead of slicing at the transient, move the transient to where you want it, then split just after during the decay of the note. transients have a tendancy to sound weird when time stretched so you should have better results doing it afterwards.

 

[PhilR]

of course you are entitled to your own ways, but as something you may want to try:

do the same method, but instead of slicing at the transient, move the transient to where you want it, then split just after during the decay of the note. transients have a tendancy to sound weird when time stretched so you should have better results doing it afterwards.

Sure it would still need to be cut in order to move it?

We're only talking about very small stretches. I haven't had any problems so far. Like I say the time stretch algo in Reaper is very good.

 

[dani]

well, recording at 88.2kHz is double the amount of information, so it's sound thinking - it may make time stretch algorithms run more accurately.

i'd give it a try, and see what happens!

i'd also bounce them to 44.1kHz after editing too, i'm sure the higher number of samples = more CPU time used by plugins..

thanks,

 

[Melodeath]

I can't imagine how it would actually help, tbh. 88.2 doesn't really have more "information," just more sample points. Whether it's 88.2 or 44.1, the audio is going to have to be re-sampled when stretched.

 

[dani]

as much as i hate to knit pick, i love it, too.. aha.

i'm not suggesting 88.2 has more audible information than a wav file at 44.1, but from a purely "information" perspective, it does contain twice the amount of information, which for temporal changes is hella important.

when using 88.2kHz as your sampling freq, the width between each sample point is halved, any interpolation done in the re-sampling stage will be twice as accurate, on average.

 

[John_C]

I've had a couple of discussions with professional engineers in person and their opinion seemed to be that the more heavy processing you're going to be doing to a sound, the higher the sample rate you need to use. I.e. classical is just fine at 44.1 but lots of heavy pitch and time shifts sounds better if everything gets done at a higher sample rate