HISE FX and delay compensation

[UD AUDIO]

This post is deleted!

[Dan Korneff]

I could use some suggestions on how to improve my approach to handling latency in my projects. I've currently been manually measuring latency and compensating for it based on which DSP nodes and oversampling are active. While this works well for static latency, I’m running into a challenge when I need to dynamically update the latency.
Right now, I’m using:
prepareToPlay() to get the initial latency with Engine.getLatencySamples()
processBlock() to set latency using setLatencySamples()
This works well for static latencies, but now that I’m introducing multiple oversampling filters, I need to adjust latency dynamically, depending on whether the user selects Linear-Phase or Minimum-Phase filtering. Since Linear-Phase filtering can add up to ~120 samples, I don’t want to set a fixed latency that high, as it would negatively impact real-time usage.
I need a way to recalculate latency when the processing mode changes. For example, when switching between oversampling filters. The goal is to update latency based on active DSP modules.
@Christoph-Hart or anyone else have suggestions on how to dynamically recalculate the latency?

[aaronventure]

@Dan-Korneff can you measure the latency of your modules, put it in consts and make the call to set the new latency to report whenever the user changes any of the controls that affect it?

[Christoph Hart]

I would not trust every host to support dynamic latency so I‘d definitely check this thoroughly.

[Dan Korneff]

@aaronventure said in HISE FX and delay compensation:

@Dan-Korneff can you measure the latency of your modules, put it in consts and make the call to set the new latency to report whenever the user changes any of the controls that affect it?

That's definitely plan B. I guess I need to use a global variable to get this data into scriptFX?

[aaronventure]

@Dan-Korneff why do you need to get it into script fx in the first place?

[Dan Korneff]

@aaronventure scriptFX is the only place you can access the processBlock to call Engine.setLatencySamples()

[Christoph Hart]

@Dan-Korneff Why do you need to access processBlock?

If you need the sample rate / block size for the calculation, attach a broadcaster to the processing specs.

[aaronventure]

@Dan-Korneff I'm calling it from a control callback where sample offset is adjusted and it works just fine (in Reaper).

[Dan Korneff]

@Christoph-Hart I was just following your Latency Test repo

[Dan Korneff]

@aaronventure Damn... Just tested and it works like you said. Happy days!

[Christoph Hart]

@Dan-Korneff ah the example might precede the broadcaster addition, so that‘s kind of outdated as a best practice but if it works…