Page 1 of 1

Looping trouble

PostPosted: Tue Mar 02, 2004 12:30 pm
by emugonzo
Hi,
I'm trying to dump some wavetable type samples to the Emu. They're 401 samples long and should loop from 0 to 400. Now if I tune the loop in Awave and SMDI dump it, it turns up beeing 408 samples long in the emu, and loop is from sample 2 to sample 401 (399 sample loop).
If I turn off the loop before dumping, the sample length is correct after SMDI-transfer. But I'm not able to set a 0-400 loop. The best I can get is 2-396, which of course sounds horrible.
Has anyone else noticed this?
Is there a workaround? I want to send approx. 500 samples over to the emu..

thanks in advance
emugonzo

PostPosted: Sat Mar 06, 2004 9:05 pm
by sampleandhold
what pitch are the samples at? if they are at 440hz, you should be able to loop the sample at, really, any number as long as i is a loop size of 100. so you shouldn't really need a sample that is 400 samples long. but i would do this anyway, because if can't cut the samples down to size because the emu will add dead space to the sample. i guess this is so it can breath.

now you can go back and find a post i did where i actualy came up with the formula for this, or you can use your calculator on the sampler to find at what lengh your loop should be depending on the lenght of the sample. of course this is all relative to your sample rate. it should be at 44.1. i can't remember if the sampler allows you to take the veriable of the sample rate in to consideration, or if it just assumes you are in 44.1.

also, if you really want to get a good sounding interpolation. you may want to sample each note at A. for example A3 is 440 hz (i have seen documents say it is actualy A4, but whatever) then sample then next up 880, then the next down 220, then down from that at 110, then of course go up to the next A above 880, that would of course be 1760. and that should be about as high as you need to go. loop points would be as follows...

110, would be 400 samples long if i remember right, next 220 would be 200 samples long, 440 would be 100, 880 would be 50 and 1760 would 25. check with the calculator to make sure.

try this, you should have it work out for you. and rememeber, it doesn't matter where your start and stop points are, just as long as they are a few samples away from the actual begainning.

now, if you really want an accurate representation, sample every other note. but really, who has time for that...

PostPosted: Sun Mar 07, 2004 5:59 pm
by emugonzo
thanks mate!
but see i've got this bunch of singlecycle wavetable samples from a sample cd. They are all 400 samples and single cycle. hence looping 100 would leave out 3/4 of the cycle.
btw why do people sample at a440. according to emu sample calc you need to loop 100.22 samples. why not use g3 @392 Hz. Giving you a loop of 225 (actually 112,5 but double it). that's what i do at least.

emugonzo

PostPosted: Tue Mar 16, 2004 4:49 am
by sampleandhold
the reason why is because western music is tuned to 440hz, A3 or A4. i have actually tried to convert my samples from 44.1 down to 440, but the loop isn't clean at that point, even though the sample lenght should be 100 flat. if you check most synths and other electronic instruments, you will see that the tuning will be set at 440 hz. my roland synth is that way.

are you sure the complete cycle is 400 samples long. that would mean the pitch is really low. i really don't know what pitch the samples are. also, you don't want to sample one cycle, the sampler isn't that exact. it adds dead space to the samples so the start and end points have room to breath. you should sample at least a half to one second lenght, then cut down after you establish your loop points.

if you want, post a sample up of one the sounds and i will give it a shot and see if i can loop it, i would like to have, like i said, a one second lenght at most.

i hope this gets you sorted.

PostPosted: Tue Mar 16, 2004 4:02 pm
by emugonzo
Thanks.
I actually never thought of the relationship between the sampling frequency (44.1kHz), tonal frequency (440Hz) and single cycle looplength (100 samples). nice!

I have solved my problem of looping. i just double the sample length and set the loop about 10 samples in and then trim the post loop samples off..

I had this strage thing happen though with one sample. looped and sounded great in mono - but after i converted it to stereo, the loop didn't sound right any more. need to investiagate that further..

Emugonzo

PostPosted: Sat Mar 20, 2004 5:03 am
by sampleandhold
i am 99 percent certian on this one: when you set your loop point in mono then convert it back to stereo the sampler fucks it up. when you convert something back... or to... stereo, what the sampler actualy does is copies the image of the wave and off sets it a few milliseconds. so you will more then likely hear clicking noises at the loop point, and you will hear it on the right channel perhaps, since that is where the "new" image will reside in. (shhh, even though mono sounds pretty much like stereo, just softer)

this is one reason to convert your breaks to mono before you go at them with your knife....

also, if your samples are all stereo you will reduce your polyphony down by half. i found that out a few nights ago. i just couldn't figure out why i was having drop outs with only 34 voices going he he he.

so the moral of the story here is:

keep your samples mono. perserve polyphony and your loop points.

also remember, you should gain your mono samples up to normalization for the loudest possible signal. if you want to be daring, do the "first six" as well. the "first six" refers to the six db of a signals loudness. the idea here is that if you boost your signal up to 0 db, you should be able to boost again by another six db. those are the transiates, the bit of the wave that doesn't matter so to speak. you can clip those and hear no distortion. you can do this easier with analog, but i have gotten away with it in the digital realm as well. remember a sine wave has no overtones (transiates) so boosting those will only make your sine sound sort of like a square wave... kind of. even at one db boost.

want to have some real fun. take a square wave and see how loud you can get it before it distorts audioly. i bet you can go a loooooooonnnnnnnggggggg way before it sounds distorted. remember a square wave is really just a really distorted sine. you can gain the hell out of saws and triangles too.

oh, i have been rambling. sorry. i hope this helps out though.