Zipper Noise

DSP, Plugin and Host development discussion.
RELATED
PRODUCTS

Post

Jeez wrote: Another way of doing it would be to have a fixed write head, with a read head of variable distance from the write head.
thats what mdsp suggested. Not the same effect as a tape delay, but a good solution.

Post

Jeez wrote: Another way of doing it would be to have a fixed write head, with a read head of variable distance from the write head. Perhaps something like:
which is VERY different.
if you use my method you emulate a TRUE tape-delay, delay length changes will sound *VERY* different from your -classic delay-implementation- method.

cheers,

- bram

Post

Jeez wrote:Well, that's just silly. You'd need some magic code to interpolate, otherwise you'd get bizzaro artifacts whenever you increase the delay length!
Aha, no, that's the whole point:
if you speed up a tape delay, you hear a pitch UP, and you even hear an echo which goes up in pitch all the time. If you've ever tweaked the knobs on a tape-delay you can hear that the "bizzaro artifacts" you talk about are actualy that what makes it fun to listen to.


- bram

Post

Bram wrote:
Jeez wrote:Well, that's just silly. You'd need some magic code to interpolate, otherwise you'd get bizzaro artifacts whenever you increase the delay length!
Aha, no, that's the whole point:
if you speed up a tape delay, you hear a pitch UP, and you even hear an echo which goes up in pitch all the time. If you've ever tweaked the knobs on a tape-delay you can hear that the "bizzaro artifacts" you talk about are actualy that what makes it fun to listen to.
As far as I know, the fixed-write variable-read method would do this too.

The bizarro artifacts I was talking about are this:

If it's a short delay, your read and write head would be skipping samples. Like you said, this would result in some positions in the buffer not being used - they'd just retain whatever they were last written with (or undefined values, if they were never written to). When you increase the delay time, you'd start using those unused positions. If your read head and write head slow down at the same time, the read head will read samples that weren't written by the write head (at least, not in that cycle).

...or maybe I'm not understanding something?

Forever,




Kim.

Post

Bram wrote:
Jeez wrote: Another way of doing it would be to have a fixed write head, with a read head of variable distance from the write head. Perhaps something like:
which is VERY different.
I realised this morning as I was waking up that I'd made a mistake in my interpolation.

Forever,




Kim.

Post

this is quite interesting:
http://arxiv.org/pdf/cs.SD/0005022.pdf

Post

texture wrote:this is quite interesting:
http://arxiv.org/pdf/cs.SD/0005022.pdf
Hoe come no-one realises that newspaper-style columns are a b!tch to read on a computer screen??? :x

Forever,




Kim.

Post Reply

Return to “DSP and Plugin Development”