Martin (or anybody) I have a great filter idea

[guyman]

Your ways are mysterious, your contributions are innumerable, and I owe you a fine steak dinner.


Thanks my comrade, I will post my related advancements.


-that guy

[guyman]

After testing:

with Q values higher than 1, and gain values greater than about 10-15 db, and the cutoff freq under 70hz, the peak filter pops into DC/blank graph...


the same thing happens with greater than moderate values in the high end frequencies, but I resolved that by limiting the cutoff freq range to values <.9 (0-1)... I would do the same with the low end, but that's really high to limit an eq...

[martinvicanek]

Yeah, that's a single-precision issue. However, at the low end you can take plain RBJ, no need for fancy matching anyway.

[guyman]

Yes very true !

[guyman]

It sometimes pops under fast modulation. Always random.

[martinvicanek]

Check your cpu load. If you update coefficients at sample rate you might hit the roof and get dropouts.
(The coefficient calculation I posted is not optimized - there is still some room for improvement.)

Other than that, if your signal has high frequency components and you modulate very fast (a few kHz or more) then, yes, you may have seemingly random clicks/pops depending on coincidences between signal and modulation variations. Ultimately that is unavoidable even for the otherwise modulation-friendly ZDF filter. I recommend to keep the modulation frequency below the characteristic filter frequency.

[guyman]

Would it do damage to have a nan/-inf DETECTOR (not the stream blocker) that sends a trigger to flush the audio streams in the event of a crash??

[martinvicanek]

Would it do damage to have a nan/-inf DETECTOR (not the stream blocker) that sends a trigger to flush the audio streams in the event of a crash??

As a last resort hack if all else fails, yes. I posted such a module earlier.

[guyman]

Thanks Martin...

This is my current solution, it pops under pressure likely for reasons you mentioned, often when flipping from the low freq filter to the high...

BUT It fades 2 filters together to get a cohesive spectral sweep....

Is there a better way to do this? ( I'M SURE )

[guyman]

oops... here


EDIT: p.s.

the envelop ranger before the matched coefficient module is set to, .04 - .09 ....... It keeps the filter from popping out when it goes lower than the crossover/filter switch point.. also I sprinkled in nan/inf blockers, seems to help -tho maybe it's placebo..

oh and the envelope ranger on top limits the cutoff <.9... the filter fails above that frequency