< Back to IRCAM Forum

Level match decorrelate~

Spat
#1

Hi there,

I’m working on a patch that generates early reflections and I would like to explore using spat5.decorrelate~. When enabled, the output appears to be significantly louder than when bypassed. Is the level difference a known quantity or is it best to measure and manually correct? Ultimately I would like to match the enabled level to the bypassed level.

Thank you,
nd

#2

I’ve found the boost to be around 33.5 to 33.6dB, but it varies depending on the filter parameters.

#3

Hi,

Thanks for the report.
I’ll try to improve that in the future.
For now, you’d have to compensate the level difference “manually”. This is not a precisely known gain, and indeed it depends on the filter parameters.

Best,
T.

#4

Thanks for the quick response! I’m working on an early reflection engine and am trying to slightly decorellate signals that are delayed between 1 and 100ms of the original signal. From looking at the exported filters, the minimum amount of decorellation smears signals over a relatively long amount of time. Is there a way to set the object so that the time smear is only over several milliseconds? If not, are there other spat objects that could be implemented similarly?

#5

Interesting questions.
I guess that reducing the number of allpass filters might “shorten” the smearing effect ? I admit I haven’t tried it. Maybe also increasing the minimum pole radius ?

The only other options I can think of :

  • generate some decorrelation filters on your own, and use them with spat5.conv~ (if they are FIR)
  • you could try out with spat5.reverb~, using very short delays and decay time. spat5.reverb~ implements a FDN, and FDNs can be good decorrelation engines.

Best,
T.