< Back to IRCAM Forum

Spat5.noisegate~ click when level crosses above threshold

Hi! I have a query about spat5.noisegate~ behaviour: the output clicks as the volume crosses above threshold, even with a slow attack time. When falling below there does not appear to be a click. Is there something I have misunderstood? You can clearly hear this behaviour in the help file by setting the cycle~ to 300Hz, adding a gain, and then changing the gain to hear the gate click on and off.

Hi Natasha,

Yep, nice catch.
I believe the issue is fixed in the attached build :
https://filesender.renater.fr/?s=download&token=a7bfbbfc-ad45-4e64-b3da-96e32ec53165

Please confirm if this is ok for you.
Best,
T.

Hi Thibaut

Not exactly. See attached maxpatch.

  1. Slowly change the gain of the input gradually below -30. Even though the threshold is set to -35 you’ll hear distortion from -30.98 to -31.16.

  2. Faster changes result in a click. e.g. alternate between -40 and -30 gain on the input.

I realise that sine tones are devil of an input test, but the click is evident on pure sounds like a soprano voice.


----------begin_max5_patcher----------
1465.3oc6Ys0aiZDE9Y6eEHTezq24JW5Ssqpz9RWoJ0G5CQqhFim3PBFrfgz
jtZyu8NW.GrCXyU6VoMODCCCbNeemuyLmC7s4yrWk7LOy15mstwZ1ruMe1L8
PpAlUb9L6srmChXY5oYukmkw1vsWXtlf+rPO9GHfxwRxEQbg3kcbyC1115qE
WZGSDbeX7laS4AByUgTukfEVHDP+i+Rp5jkf82Sb91vX4CTaeTwfgq0VMY0C
e.Y+1DMlVOSnZvuOet5eKZI3tKJQ9PZFGKrrWwh2bR7PM.ARIpenFTUEO2kj
tkomrScHD9dDhcs2auT1VtfmdKOlsJR6XfZgO5JA+xvIEMdvmbYfeSRaTOU1
NtZN.hv8UYSGOoci4sN8DcTughN3E.cX5xdBOvPgGX5gGsmwNLbffCNhJyX9
eKejuCaQIr05UaZBgmconBT5nWCpbWll.YMq6.wWBAJnmBTeWMfHt8MDhldz
8wMrvXqeBNHD5Y1HEd53WMHj7+.MpAiPOu9pQmbLlI8Z5R8ttu1HNyB2DyhT
EHbRzhP5XIVJWk+352c35WOdKi9Yrm3quUNSoIukIDogqxElBbmsG1MTGg9Z
ZNajXt0rfWsfVH6SrDEfp3.B1nwocViC8pmP.iGLBdIHh+pzUAmK7WoFuXQr
jg02+ulFpjFmSTTjm6XxygcfCv6YXkcyB+GscgpGzDmaHtOMuEqs0I5.CML.
1oXg8SRG0UtLbxoiFWuOkuNOPDlzxE86DwTVvomYwCRm0IH2qkNIiscWlHYa
VaVAU1Elp+ndwQksbPaEGUi3gzsbowrMqOJyl3Y2mDsdJDOfAJdnWyjpHNKi
OArB1uJq3P6Lq3c8XE4t6rfGmBRAVkTJpIqKjh+0iThRRdjcOmMEoPkqsX3E
OP2W+kb0V+UWCabRXFeCSHKl4WBtmEGyixdq5rSrhbcU21m8zwNZwDzo6bGs
abGdRpGV6s1QgwG+Ji0PVM9gglrj7zfR5rHyv5MLulmIBiYpxEpLGU4KUlz1
v06RBiEEVDKYMcmtEMR.1el0WqUuzVWyavtFALQtlJ1eVeCdH.RRWySaNKZ7
MM78lFLLSO3.BENQADUQRU.b89FhbJeyUVdOUsXJR8iiGb+YC02ZS.qHObPF
R8QINqkniggZCjHiggPWLtC1FtaLvDxc3oQ9STZDpMo3DzXvBjAyBEonS.Kz
FsPYXbXVBbgxXaSdDdLT2P+KUdDrMZU+oXK3VY4F17eX6.q99GmWYRudlFNI
7cqLMndSildSCafvwCyzn1uI6Hy2cY0hQVk0F5986+XZChsa2S7zrhYqsgrW
zGRz9k2B8ogwlS0MzYmxeJrb9lQXoxNEExdyxSMMR9rioOO6sIRDFmGVvuRz
IMYl3knia85sdPyWGl7mBlHO61uviKduzRTcGKORbHSrZycgQQAIQF2qZqfk
M.aat5hhg2O2ar.KQ9DHzSRKKwPrKxQej7.JsRYwE2Cr7lHThO.olJwA4Rn5
i7PXJVcD3naC8ls.PeiE.9d.h4H4PPospdar3MlFWQtfxthkRgzjcIokwT4s
5ue94hjMor0g7Xw662cQgPJUdUMsq4jBNcqjPCKumpBu8Aieyv5xFz+j58Y9
GIpWsP3oBIUaq+Fci8kZtQxU9B42Om8KumaJdiFVJm+vW4wwtX024gJi6HcR
ywNvRO53iv1xzcCof+Chzxu7a+Sv2K9qOeE65Pb0tO1ifgZHQNJmqmopsl55
UBdQdbkP4crxk8g0996JUAmVpOtQP1ytsQr9i.43rf7CqV8CZts4KsMuvezR
KdLKJTVvyeopF5TwoJLg9Mpev+FIew7MD9TTN+zqwxBBjF3fXCFpDApHhy9p
c7KzEisC94TNOtydHP4RTr5O0QPGOYwXipykucEO8yxsS6ruoHJfVYS8QPpl
GgUU1c2AMkpq+9PJOLaWwNA5Oiz7uO+eAEP.nZB
-----------end_max5_patcher-----------

Hum…

  1. Due to the running mean amplitude, the gate is indeed triggered when the input gain is ~ -31 dB. (the rightmost outlet reports the gate activity)
  2. Here is what I get when alternating between -40 and -30 dB input gain. (other settings similar to your patch)

    No noticeable artefact, isn’t it ?
    Are you sure you updated the spat5 package ?

Any specific sampling rate/vector size to reproduce the issue ? Or can you send a specific sound sample ?

Best,
T.

Hi!

I checked the package info and its says “version” : “5.2.8”.

Attached are two examples illustrating the two observations. Changing sample rate and i/o vector makes no difference.


I’m stumped as to why your example is clean and mine isn’t!

Natasha

Maybe I mistakenly sent you a wrong package.
I’ll send a new fresh build later today.

1 Like

New attempt :

https://filesender.renater.fr/?s=download&token=c0fd8532-7dc0-4144-b873-c5faf35b7daf

Yep, that one works. Thanks,
Natasha

Excellent ! Thanks for your helpful feedback.

T.