seekyou

Delay element in gen accept int as samples delay. What happens if a float is fed into the delay block?

Is there any element which can accept non integer values as delay (tapin/tapout ?)

delay-and-tapin

Depends on what the interpolation mode is for the delay. If it's step, then truncation (or nearest? Worth checking!). There are several interpolation modes available as well, so those will be fine with floating point values.

Depends on what the interpolation mode is for the delay.  If it's step, then truncation (or nearest?  Worth checking!).  There are several interpolation modes available as well, so those will be fine with floating point values.


What I am looking for is an interpolation that won't introduce any change in spectrum amplitude. Where can I find the possible interpolation modes for delay block?

It sounds like you want allpass interpolation. That's not an option in gen~, but you can find code to do it. IIRC the interpolation modes available for delay are the same as those available for waveforms. (step, linear, cubic, cosine, spline, IIRC)

Oli Larkin has posted an example of allpass interpolation on the forum. You can also find the standard algorithm for allpass delays online; it's pretty straightforward.

It sounds like you want allpass interpolation.  That's not an option in gen~, but you can find code to do it.  IIRC the interpolation modes available for delay are the same as those available for waveforms.  (step, linear, cubic, cosine, spline, IIRC)

Oli Larkin has posted an example of allpass interpolation on the forum.  You can also find the standard algorithm for allpass delays online; it's pretty straightforward.


You mean I can design my allpass filter with the phase correction I want and include that in my delay line patch?

Will I have to rewrite the entire thing in a subpatch or could I continue to implement it in gen~?

You'll have to implement this one in gen~, but it's easy enough to figure out from the block diagram.

Good, that's what I wanted to do in the first place.

I know how to do it in matlab, it's a 1st order all pass. It defined by a single coefficient.

so from this I can derive the block diagram in gen

Does this look right for the previous equation of the all pass filter?

My in3 is the C parameter. The allpass is the second section of this loop filter.

Does this look right for the previous equation of the all pass filter?
My in3 is the C parameter. The allpass is the second section of this loop filter.


delay and tapin