Whats up with all this glitchyness going on?

Ok so after I pretty much got over my pads being a little off the OS is starting to drive me mad. I tried the slice samples option today. I knew my sample loop was 109bpm as the full sample loops fine in the sequencer set to that. So then I go to slice and set it for 16. I went through and edited the start and end times so I had nice clean cuts. Sequenced it out at the same bpm and noticed some choppyness.

Now know that altering start times of the sample chop points and then entering in perfect 16ths notes will result in a little timing discrepancies but this seemed more than normal (keep in mind I've been chopping loops and working with samples for about 16 years now, I'm not new to it by any means). So then I go and rechop it to 16 slices but this time keep the start times unchanged. I then sequence out 16th notes and play it back. Now playing back the sequenced chops at the same tempo as the original should result in a seamless loop.

Doesnt matter if the transients line up perfectly or not. Your chops are even and your sequenced notes are perfect as well so it should just play along like nothings changed. Its not till you change the tempo that you should hear anything. Yet my loop is sounding like all the parts are too short.

I double and triple check everythign, redo it a bunch of times and no avail.

I then take some drum loops that are perfect drum machine tempo and chop then. Some of those fare better but others have weird little pauses in them. Again this is all playing back at the same exact temp as the original. You should hear no difference.

So I figure ok maybe something is wrong with the slicer, I can always do it manually in a DAW if need be like I normally do now.

So then I'm preparing for a live show and I've got some long pad sounds I need sequenced. I export them from Cubase as 8-16 bar loops. Thats when I notice them being cut off. So then I read adivce here that says to chop all your stuff up in smaller sections so it doesnt get cut off (just so you know the polyphony is not a problem). So I try using the patch phrase as well as slice to pad. Both options give very audible glitches. As a last resort I slice in Cubase and load, same thing.

What the hell is going on with this thing? It seems to not be able to play a sample exactly as its timmed. There is some delay in the start of the samples. I've gone to the AMP envelope to try and tweak it there but no luck.

Don't get me wring for sequncing drums and shorter sounds I love it. But this is a serious problem here.

Is there anyone able to play samples longer than 8 bars? So far that seems the limit. There seems to be a bug where its cycling/reseting its voices every 8 bars or so. Very frustrating. Can't remember when a piece of gear got to me like this. I've always been a fan of Akai. Have used the S612, S950, S1000, S3000, MPC-3000, S5000, and now the MPC-1000.

Come to think of it though they all pretty much had their probs. Even the MPC-3000 would glitch. I had to spend hours sometimes rewidnign my DAT to get a good take from it. And don't get me started on the S5000. I foolishly bought one right when it came out...

Anyways if anyone can offer a solution I'd really apreciate it. At this point it looks like I might have to run a second sampler/computer for the longer samples which is pretty stupid.

Sorry for the long rant...

Oh and just a little experiment to show that this has to be a bug and not valid note stealing.

If I have my pad playing with my shakers (along with the rest of the drums) it will cut off at bar 5. If I mute the shakers then it will play past that and if I unmute the shakers right after bar 5 it keeps playing. If this was a polyphony limit problem then it would cut out as soon as the shakers came in but it doesnt.

But once it gets to measure 9 then it does cut out.

So there is no problem with it playing all these notes as once. Sometime is happening at the top of the bar that is making it cancel that voice.

Basically no new notes are encountered to make the pad cutoff, it just decides to for whatever reason. I basically have the same amount of loops running for the entire 16 bar loop. The amount of voices is constant.

If only I could chop this to get around it. Might try adding some release to the sample decaying and a little attack to the one coming in to see if I can blend the glitch out.

Well luckily I got patch phrase to sound ok on this pad. Chopped it to 4 parts (16 bars total) and then used a decay of 70. Since its a heavily effected pad in the first place the slight coloration from the decay parameter (timestretching the ends basically) isn't noticable.

Deffinatly a unecessary workout but I'll take it at 5:30 am


Hopefully Akai fixes this but for now it seems like I can deal with it.

For making beats in the studio I love this box, but I got it for live so its got to hold up in that respect.

Doesnt matter if the transients line up perfectly or not. Your chops are even and your sequenced notes are perfect as well so it should just play along like nothings changed. Its not till you change the tempo that you should hear anything. Yet my loop is sounding like all the parts are too short.

sounds like your program is in mono. go to program, param, and in the top right hand corner set it to poly. are they cutting each other off? whenever you chop to new pgm, the pgm created is automatically a mono pgm...keep that in mind.

PLEASE, people, if you have had your mpc for under a month, PLEASE PLEASE PLEASE keep the word "Bug" out of your mouth

It might havebeen in mono but it shouldnt be cutting off anything.

If you divide any sample up to 16 even parts and then play those back at the same tempo as the original it should sound like the original loop. If you slow down you get gaps of course and if you speed up then the chops will overlap.

Whats weird is that its random. I've had some loops come out fine, others have the first chop with a gap and so on.

I'm going to experiement more. I'll try chopping in a DAW and then playing it back.

My only clue is that the loops that had the glitches were ones where the 1k couldnt detect the correct tempo (I knew what the tempo was regardless but I was testing it).

Anyways I'm glad ot see your positivity on here Nym, gives me hope that I will eventually sort all this stuff out.

gives me hope that I will eventually sort all this stuff out.

without a doubt, you will. there's always a workaround, solution, or w/e.

hey here's an idea - what are the kinds of samples that you chop to your satisfaction, and what kind of samples sound choppy?

Just did a test and it came out fine so i at least see that it can work right.

The ones that were choppy ranged from a sloppy break beat to a perfectly quantized 909 ride sample. With the 909 one I chopped it to 8ths and the transition from the 1st chop to the second had a glitch but the rest were fine. This was using patch phrase. The sloppy breakbeat just sounded like all the samples were too short, had that gated feel.

My worry was that the 1000 wasnt capable of playing bacl samples properly but looks like it was just a fluke or maybe some error on my part. I'm sure over time I'll sort it out. Just panicked a bit I guess.

Right now the main prob seems to be when I'm doing a live show I'm basically bouncing out 4/8/16 bar loops of my synth patterns to play back in the mpc. For stuff like long pads you can hear the chops (which are necesary since it seems to want to cut them off at bar 5 or bar 9). It seems like the decay feature on the chops will help using patch phrase to blend the glitch from going to one slice to the next.

Guess I have to realize a sampler isn't going to be sample accuate like a DAW so anything slightly off on a sustaining sound like a pad is going to be noticable. Wondering if I should call Akai about the cut off note problem...

Anyways thanks again.

Ok I basically figured out the problem.

Whats happening is that when you tell it to chop in even pieces first of all its losing the tail end of the sample. So in the case of the sloppy loop I didn't go and edit the start points, even though normally you would. Well I did the first time but then when that sounded weird I tried just a regular straight up even slice. As long as you play back at the original tempo it should sound ok, which it didnt.

So I see what happened (when I took the slices into SX) is that one of the snare hits was early (slice 4). So the beginning part of the snare transient was at the tail end of slice 3. Well as long as they play together it will sound complete. Problem is that when I looked at slice 3 there was no snare transient at the end. Now when I was in slice mode I saw it there. Yet when I converted to slices it was gone. In fact all the ends of the slices had the sample data basically faded out.

I experimented and found that if you "extract" the slices after you do the slice function then they come out ok. So in this case it preserved the beginning of the snare transient on slice 3. Some how the "convert to slices" doesnt work right. So work around is to just extract them.

I have a link here to a pic showing whats going on. Notice the red circles. The top one shows the original sample and see how there is a constant low level signal right up to the main transient. Then see on the bottom example how the converted slices look. You see that first of all there is a gap, so the sample is shorter (I verified this by looking at the length of the slice before converting and aferwards, they dont match, the converted was shorter in this case, decay was set to 0, but 0 should mean it comes out normal). You can set the decay param higher to get rid of the gap but you still wont get the end of your sample, it just timestretches the end but it still misses out the data that should be there.

You can also see that the timing is off. No idea whats going on here but the transient should be in the same exact place. All I did here was slice it then copy to my computer. It shouldnt be moving the sample data around. All it should do it just divide it into even pieces.

It almost looks like they are trying to do some sort of transient detection. Maybe it is, that would explain it moving samples around. But we still need the ends of out slices preserved.

So in some cases you wont notice the gaps and as long as you have clean transient starts you wont be missing the ends of the slices before them. but even when I went in and edited my transient starts on the slices before converting it still sounded weird on some loops.

Note that the loop in this pic is actually one that sounded ok, its not the one I'm referring to up top.


So workaround for me is to use recycle or divide in SX.

hope this helps someone

www.medwaystudios.com/media/mpc.jpg

Heres a pic of the transient thing I was talking about.


This is basically the difference between using convert or extract after you make the slices. Convert is basically screwing the data up. Youll see the one slice on top is the converted one. Notice the lack of transient near the end. Its longer becuase I tried adding some decay to see if that would bring it back but it doesnt. Akai must think we want the samples cleaned up at the end of the slices. In some cases it might work but I'd rather it be left alone.

The bottom slice is what it should look like, again this is using the extract function. Notice how not only is the beginning of the snare transient preserved (at the end of the file) but the intial transient at the start is different. The top slice looks weaker. I noticed this intially when I'd use convert, all the transients looked thinner.

So altho the convert feature is time saving in the end I'd stay away from it as I don't want my slices altered. Wish Akai would at least mention this in the manual. Would be cool if they had a purer convert function that didn't alter the slices as well.

www.medwaystudios.com/media/transient.jpg

medway wrote: Its longer becuase I tried adding some decay to see if that would bring it back but it doesnt.

That's because you are misunderstanding the Decay function. Do not think of the decay as a typical decay like you would find on a synth or a reverb plugin. The sample decay (and the decay setting when you convert) is really not decay, it is more of a fade out, so it will not "extend" anything, it is simply the length of the fadeout.

Soooo... the reason the transient did not show up on the first waveform is that you did not have the decay set to zero. Adding decay is actually making it start fading out sooner.

I hope i am correctly understanding what you did.

TBonus76 wrote:

medway wrote: Its longer becuase I tried adding some decay to see if that would bring it back but it doesnt.

That's because you are misunderstanding the Decay function. Do not think of the decay as a typical decay like you would find on a synth or a reverb plugin. The sample decay (and the decay setting when you convert) is really not decay, it is more of a fade out, so it will not "extend" anything, it is simply the length of the fadeout.

Soooo... the reason the transient did not show up on the first waveform is that you did not have the decay set to zero. Adding decay is actually making it start fading out sooner.

I hope i am correctly understanding what you did.

I do understand it. I made a typo tho, I meant to say the "release" parameter. I was refering to the parameter that comes up in the slice/convert dialog, not the program settings. Remember this picture is from a direct save of the sample, not it being played in the MPC. This is what the convert function destructively (ie permanantly) does to your raw sample before it even hits the AMP env.

Also the decay in the mpc can act like a reverb decay (called RELEASE on synths) if you set it to END and you let go of the sample before it reaches the end of it (with NOTE ON set). It can also act like a DECAY on a synth which fades down to the SUSTAIN level (which on the MPC is fixed at 0) if you set it to START. Its basically one or the other depending on where its set to START or END.

But yes the Release parameter in the convert function will extend the sample, it does not fade (well it does end end up fading a bit but it primarily extends). Check page 62 of the 2.0 manual and you'll see what I mean.

Like I said before setting to 0 leaves the slice with gaps and missing data, adding to the release extends the end of the sample (through timestretching which is obvious if you ever really listened to it at any high setting) but still leaves out the data.

Plus its also obvious from the pic that the sample got extended and yet also missed the transient. Did you notice how the top slice was longer? Yet you tell me release doesnt extend it? Look again...

Ok, I see... I was confused when you said decay, and I forgot about the DECAY/END/NOTE ON thing. Thanks for clarifying.

Glad you brought it to attention because now it makes more sense to me that it's labeled release.

Antonym wrote:PLEASE, people, if you have had your mpc for under a month, PLEASE PLEASE PLEASE keep the word "Bug" out of your mouth

word

TheFirstJimBo wrote:

Antonym wrote:PLEASE, people, if you have had your mpc for under a month, PLEASE PLEASE PLEASE keep the word "Bug" out of your mouth

word

I wont call this bug, Akai probably considers it a feature.