Tremolo effect

[RJ Skerry-Ryan]

We should find a nice FLOSS gater effect or write one.

[Phillip Whelan (pwhelan)]

There is a great one that was implemented for Buzz Machines: https://github.com/Buzztrax/buzzmachines/blob/master/Arguru/GoaSlicer/goaslicer.cpp.

Unfortunately I was unable to find a license that applies to it. I remember though that when I tried to implement it I needed to know the current BPM and the sample(s) at which beat boundaries occur.

[RJ Skerry-Ryan (rryan)]

Kicking to 1.13.0 unless there is a candidate for 1.12.0 that emerges soon!

[Johan Lasperas (johan-lasperas)]

I started looking into implementing this feature, I think it should be manageable and fun enough to begin being involved in the development.
As I see it, applying a (smoothed) square envelope on the signal is a good solution as a first shot. Currently testing this, with a rate & a shape (=duty cycle) parameters.

[Be (be.ing)]

Cool, thanks for taking this up! If you'd like feedback before it is finished, open a pull request on GitHub and note in the description that it's a work in progress. I suspect that following the example of the Echo effect's Time parameter for this effect's Rate parameter would make sense. The Echo effect's Time parameter has a short range with a linear scale rounded to the nearest 1/4 beat. The code of the Echo effect is here:
https://github.com/mixxxdj/mixxx/blob/master/src/effects/native/echoeffect.cpp#L144

[Johan Lasperas (johan-lasperas)]

I've got something rather straightforward and sounding good by implementing a simplified noise gate :

A trigger is generated depending on the rate parameter (1/8, 1/4, 1/2 or 1 beat right now, can be extended to free rate, or patterns)

The gate outputs an Attack-Hold-Release linear gain curve when triggered, applied to the signal.
Hold time is given by a shape parameter knob, while attack & release slope depend on the rate.

While playing around, I realized that when many channels are active on the same effect group, they are all processed serially (numSamples Ch1, then numSamples Ch2, then numSamples Master, ..), which is annoying when trying to keep the effect group State consistent between two buffers of the same channel.
A simple example is to enable a time-dependent effect (echo for example) on Ch1, then to enable a second channel on the same effect (Master or a playing deck on Ch2) : the rate is doubled.

One solution would be to keep a state per channel in the GroupState (in this case : channel gains, channel gate states, ..), and use the given channel handle in processChannel to access them, but I haven't seen any other effect implementation keeping channels independent this way (the ChannelHandle is always disregarded).

Any thoughts on that ?

[Be (be.ing)]

Keeping state per channel is exactly what the GroupState is for. Add whatever variables you need to the GroupState.

[Be (be.ing)]

I'm curious to play with this effect placed in a chain before Echo :)

[Johan Lasperas (johan-lasperas)]

Pull request made : https://github.com/mixxxdj/mixxx/pull/1370

[Swiftb0y (swiftb0y)]

Mixxx now uses GitHub for bug tracking. This bug has been migrated to:
https://github.com/mixxxdj/mixxx/issues/7386