Sat Jan 16, 2021 11:33 am
I'd support this, but not sure how complex it is.
If possible I'd like to add some extra complexity!
The sidechain signal is given a threshold before it activates the reduction in the other channels. This stops random noise triggering the reduction. (Including breathing when not talking, assuming no heavy breathing.)
The difficulty is that a genuine voice signal starts below the threshold in volume, and until it reaches the threshold the other signals are at full volume and the start of utterances can be lost.
It is of course impossible for a circuit to see the future and trigger in anticipation. What is possible is to delay the output signal and then trigger the reduction based on the undelayed input, so the length of the time of the delay benefits from the reduction.
We of course don't want to delay audio just for this. However sometimes we are delaying the input audio anyway in order to be in sync with delayed video. So if we are doing this, can we have the option to put in an anticpation setting for the sidechain up to this amount of delay?
(I'm not sure if Resolve is doing this - it might be able to by reading ahead in the audio track data, since it is working with recordings, not live.)
Thanks,
Tim