slice problems

Well, crap.

It looks like it's *not* something I'm doing wrong, this is just how the frickin' new "slice" function works. I mean, I can even see *why* they did it this way; it makes it easier for folks to cut up beats without hearing clicking and popping...

I noticed on Tuesday that I got very similar results with the split function regardless whether I created a new program or a patched phrase.

So I decided to do some more tests tonight with the "patched phrase" function. I rendered off a couple of wave files, one of white noise, and one of a pure sinewave, using the function generators in Wavelab. I loaded these waves into my MPC1000, split them 32 times using the split feature...

Anyway, here's my results:

http://mux.ca/mp3/testwithnoise.mp3
http://mux.ca/mp3/testwithsine.mp3

As you can hear, each "split" is quite audible. Examining the resulting wave file, it appears they ramp the volume down to zero for approximately 5ms, then ramps it back up even quicker.

This makes sense for drum patterns, where each percussive beat is likely to fall on or near one of the splits, and this technique will save a lot of people from frustration with clicks and pops... but I really wish they'd also provided a way to turn it *off*. In fact, Akai, if you're listening, please consider that an official request!

Basically, this means that the "slice" function I've waited so long for is not the feature I thought it was. It is not acceptable for use on sustained tones like basslines.

Please, someone prove me wrong!!

hey,
do you think you would get better results using Recycle or similar plus Bluebox to load up the programs?

i realize you were hoping to do it in the mpc tho..
but the software method could potentially be really fast also, especially since your loops start in ableton live.

sleepersriddle wrote:hey,
do you think you would get better results using Recycle or similar plus Bluebox to load up the programs?

i realize you were hoping to do it in the mpc tho..
but the software method could potentially be really fast also, especially since your loops start in ableton live.

Hmm - ok, just tried that, actually, with an old copy of Recycle. Interesting; I'm also getting clicks and pops, when I set a grid of 16 and did "add slices on grid"... so maybe there's something I'm just not getting (I've never really worked with loops before now )...

Hey, is that Graeme, btw?

BUSTED!!



maybe it is but don't tell anyone

we should do the soma thing again soon

Sounds like you need to be more careful to chop on the 0 axis points like someone said. I'm not familiar with ableton, but if it has an auto fade for start/endpoints of chops I would set it to about 5ms, that will guarantee that the start and ends of the chops fade to a 0 crossing.

TBonus76 wrote:Sounds like you need to be more careful to chop on the 0 axis points like someone said. I'm not familiar with ableton, but if it has an auto fade for start/endpoints of chops I would set it to about 5ms, that will guarantee that the start and ends of the chops fade to a 0 crossing.

Thanks, but all the chopping is being done on the MPC using the new "slice" feature, and that's exactly what it's doing. That's what I'm complaining about... you can hear the auto fade, and I really don't want to - it makes the new slice feature pretty much useless for loops of stuff like horns, vocals, guitars, anything non-percussive.

i chop that kind of stuff and never notice any autofading...

autofading would SUCK, there's no point in it - if akai did implement it they put their feet right in their mouth

Antonym wrote:i chop that kind of stuff and never notice any autofading...

autofading would SUCK, there's no point in it - if akai did implement it they put their feet right in their mouth

Can you try chopping up a pure sinewave for me, and see if you notice it?

I'd be so pleased if this is just caused by a setting somewhere...

i could be missing something, but maybe adding a little release to the volume envelope (in the program) will smooth out the clicks.

yo mux, after thinking about this issue more i came to this conclusion:

Say you have a long smooth sample you want to play back. If you want to play it faster than original, there is NO 'perfect' way to do that without causing some kind of artifact. (Basically you have to crossfade between the different chunks, whether at the large 'slicing' scale or at the small 'granular' scale. Or else just retune but that's not what we want.)

Likewise, if you want to play it slower, you have to somehow compensate for the gaps between slices. eg. with some stuttering, or some smearing or crossfading.

Thus, the only artifact-free way to get your original sound using the slices is to play them end-to-end at the EXACT original tempo. The only way to attempt this apparently is to use EDIT->EXTRACT from the slice menu, rather than CONVERT. (Because the waveforms thus produced look non-faded, to me.) Or, to use recyle etc. At this point, the last step would be to get the 1k to play each slice in order, at the perfect times.

But, is the 1k sample accurate? There's reason to believe it ain't. People have remarked you can sometimes hear varying phasing if you trigger duplicates of the same sample at the same time, which proves this point. (I want to still test this myself using a software editor to create sample-accurate slices at a known tempo.)

Thus, artifacts are unavoidable when playing a sustain-y sliced sample.

Which leads me to a suggested workaround::
Make your sliced program on pads A1-A16. Then just put a copy of the original sample on B1. Make it some type of mono so that hitting an A pad cuts the B sample.

So, when you want to hear the non-artifacted original, play B1. When you want to rearrange or mangle the loop, use A1-A16. Once you're mangling things, perhaps the brief fade-ins are acceptable.

One last question though:: has anyone confirmed if it's the RELEASE setting which affects that automatic fade -in/-out that Mux has observed? If so, are you saying that a zero setting still leads to some fading?

sleepersriddle wrote:Which leads me to a suggested workaround::
Make your sliced program on pads A1-A16. Then just put a copy of the original sample on B1. Make it some type of mono so that hitting an A pad cuts the B sample.

So, when you want to hear the non-artifacted original, play B1. When you want to rearrange or mangle the loop, use A1-A16. Once you're mangling things, perhaps the brief fade-ins are acceptable.

OH HEY! There we go, that might just work! It's still a lot of work but hey, that's what livepa is all about.

One last question though:: has anyone confirmed if it's the RELEASE setting which affects that automatic fade -in/-out that Mux has observed? If so, are you saying that a zero setting still leads to some fading?

I'm not sure - I tried a lot of different variations, from a RELEASE of 5 thru 40 - longer releases definitely leave more artifacts in the playback of the loop, but shorter releases (even 0) still have the same fading.

The samples using the sinewave and noise that I linked above are actually made as a patched phrase, so there's a chance there's a default release setting that is used for patched phrases. It's not a selectable option, at least.

Hey there Mux,

Sleepersriddle is on it - I was trying to elude to this before. No biggie, you don't know me that well, except for on livepa.org. Anyway, trust me - I have done a TON of slicing up loops, and I can only reiterate that your experiment - while educational - will only prove frustrating to you. A 4-5 ms ramp up/down is a common technique to deal with the problem of the zero crossing dilemma. (Akai, he's right - this should be a toggle!) Even Ableton Live's clip editor has the FADE botton for this very problem.

You're right in your analysis that percussive material lends itself much better to this process. A low freq sine wave is going to be the most non-compatible material of all, since you will be faced with much larger intervals between each zero crossing. White Noise is also problematic, but you could nail down the zero crossing with extra precision. The Akai's LCD view of the wave is obviously not as accurate as wavelab's (my fav too.) Furthermore, you can snap to zero crossing with most software editors.

A couple of tips: stick with material which has distinctive attacks in the waveform - drums, arpegiations, rhythmic bass and synthlines.

Also, moving into higher tempos than the phrase's original, will usually yield better results. This works out for me really well since I am into breaks/DnB which usually is at higher tempos then the original source material.

I would concur with the suggestions that you use software to do your preparation of the loops in question. You will achieve a higher degree of accuracy, and once the work is done - it's done. Akai was very smart to implement the patched phrase functionality, but the built-in slice editor can't compete with the precision of something like Recycle, orbetter yet - hot tip - Phatmatik (I love it, and you can create the various groove templates through midi.) http://www.glaresoft.com/purchase.html#pmp