Re: Xfade vs. zeroX / HW vs SW (was: dream box)

["Dennis W. Leas" <dennis@mdbs.com>]

>. . .
> So as I undertood your picture, I am the father of the EDP denomination.
> And I feel that this dificults my participation / contribution in the
> others, just like in real religions: If you are strong in one tribe
> the members of the others keep distance.
> . . .

As with real religions, we must concentrate on unity and not division!  And
celebrate different views as additions.  Like in physics, "Is light a wave 
or a
particle?"  Embracing both views increases our understanding.  Holding one 
view
is limiting.

>. . .
> shure, the click goes away, but couldnt you hear some bump, like as
> if an new such hum tone would start? I had that a lot when editing
> sound files on the Mac.
> The only way to get rid of it is a slow Xfade.
> The savest place to cut is an attack. I started cutting tape: Roll
> backwards and forewards until you locate the beginning of a note and
> cut there. Shure, the diagonal tape splice also corresponds to a
> cross fade...
>
> So instead of searching for a zero-crossing, you may search for an
> attack. Usually the musician starts a loop with a new note :-)
> Cutting at the last zero cross before the attack, you probably dont
> have to care about the end point, because the click will be
> overrolled by the attack, right?
>
> If there is no attack, we need a Xfade, and it can be a long one.

It probably does depends on the audio material.  I haven't heard a click 
or bump
when I test with humming, but I haven't tried many other kinds of audio 
yet.
Perhaps we could trade small sound files?  That way we could hear/see the
different atrifacts.  For example, I could send you before and after 
files.  If
you're interested, I'd suggest an uncompressed format like AIFF or WAV.

BTW: How slow is the Xfade?

> >My zero-crossing adjuster alters the loop length, even if by only a
> >tiny amount.
>>. . .
> In the worst case, you have to go back almost a phase of say a 33Hz
> note, which is 30ms. If it happens on both ends, its 60... couldnt
> you correct backwards on both ends, so you can do it in real time and
> the error would subtract instead of adding?
> With the attack method, you have to correct more, maybe, but usually
> only the entry point. The original loop length can stay untouched,
> unless there is exactly another attack at the end, which is quite
> likely. But in this case you can move the end point to the attack and
> hope that its played more acurately than the musicians foot action,
> so you actually improve loop length.

Yes, I'm concerned about the low tones.  60 ms gets significant!  I like 
the
idea of "correcting" the same direction on both ends so as to minimize the
length changes.  Seems like the best action is to quantize the recording to
zero-crossing points.  Right now, my zero-crossing adjuster works on sound
already recorded into wavetable memory.

> Ask IVL, they must know a lot about this stuff!

Ah!  The Electrix / IVL folks!

Hey Damon!  How does the Repeater handle loop splice points?  
Zero-crossing?
Cross-fade?  Both?  Other?  Adjustable?

[ I'll cc Damon this message.  Hope you don't mind, Damon.  I'm really
interested in how the Repeater does this. ]

Dennis Leas
-----------------------------
dennis@mdbs.com


----- Original Message -----
From: "Matthias Grob" <matthias@grob.org>
To: <Loopers-Delight@loopers-delight.com>
Sent: Friday, September 15, 2000 12:11 PM
Subject: Re: Xfade vs. zeroX / HW vs SW (was: dream box)


> >  >>I think both Kyma and MAX/MSP provide solutions.  This is cool!
> >  >>[Sometimes I feel that, if I'm into this "looper religion" thing,
> >then Kyma is like my denomination. :)  That makes the MAX/MSP folks
> >(and Orville users, etc.) like a different denomination; same
> >religion, they just use different words to mean about the same
> >thing!]
> >
> >>please let me be member of such denomination!
> >
> >I always thought you were one of the founders!  One of the "fathers of 
>the
> >revolution"!!!  Viva la Loop!!!!!!  :)
>
> hmm, I hope it will not turn into revolution, but a peacefull evolution. 
>:-)
> I contributed with the currage to launch the first dedicated loop unit.
> But looping has a much longer story.
> So as I undertood your picture, I am the father of the EDP denomination.
> And I feel that this dificults my participation / contribution in the
> others, just like in real religions: If you are strong in one tribe
> the members of the others keep distance.
> The SW looper specialists may fear that I grab all ideas while in
> fact the oposit is happening lately!
>
> >  >Doesnt the audibility depend on the sound material?
> >>Roughly: For percussive sound, the zero crossing is great but for 
>sustained
> >  >sound, cross fade is necessary, otherwhise you hear a new attack 
>which can
> >  >be about as annoying as a click.
> >  >In a future HW solution such fades will be available and probably
> >controllable.
> >>The sound material could be analyzed to define characteristic.
> >
> >Interesting...
> >Yes, I think the audibility depends on the sound material, but I
> >would say just
> >the opposite!  I.e., I'd use zeroX for sustained sounds.  Here's my
> >experience:
> >I wrote a zero-crossing adjuster for Kyma recently.  To test it, I 
>created a
> >short loop by humming a continuous tone into the mic and punching
> >in/out.  Sure
> >enough, I heard a click at the loop point.  After being
> >zero-adjusted, the click
> >completely disappeared.  But perhaps with other sounds, the zeroX would 
>leave
> >artifacts.
>
> shure, the click goes away, but couldnt you hear some bump, like as
> if an new such hum tone would start? I had that a lot when editing
> sound files on the Mac.
> The only way to get rid of it is a slow Xfade.
> The savest place to cut is an attack. I started cutting tape: Roll
> backwards and forewards until you locate the beginning of a note and
> cut there. Shure, the diagonal tape splice also corresponds to a
> cross fade...
>
> So instead of searching for a zero-crossing, you may search for an
> attack. Usually the musician starts a loop with a new note :-)
> Cutting at the last zero cross before the attack, you probably dont
> have to care about the end point, because the click will be
> overrolled by the attack, right?
>
> If there is no attack, we need a Xfade, and it can be a long one.
>
> >My zero-crossing adjuster alters the loop length, even if by only a
> >tiny amount.
> >It advances the start point and retreats the end point until a 
>zero-crossing
> >splice is achieved (with the same signal slope on each side).  The nice 
>thing
> >about Xfade is that the loop length can remain EXACTLY the same.  In 
>thinking
> >about Alex's original question some more, I think maybe a cross-fade 
>for the
> >sub-loop to eliminate clicks but trim the new material with a 
>zero-crossing
> >adjustment.
>
> In the worst case, you have to go back almost a phase of say a 33Hz
> note, which is 30ms. If it happens on both ends, its 60... couldnt
> you correct backwards on both ends, so you can do it in real time and
> the error would subtract instead of adding?
> With the attack method, you have to correct more, maybe, but usually
> only the entry point. The original loop length can stay untouched,
> unless there is exactly another attack at the end, which is quite
> likely. But in this case you can move the end point to the attack and
> hope that its played more acurately than the musicians foot action,
> so you actually improve loop length.
>
> Ask IVL, they must know a lot about this stuff!
>
>
>           ---> http://Matthias.Grob.org
>