Please could you try this one?

https://www.expert-sleepers.co.uk/downl ... _2_8_4.zip

I was planning to work on it next week.

Your best bet would probably be to calibrate with another well-behaved VCO and then re-use the calibration for the Plonk.

It means 8 stereo voices.

Track at 0dB (and stereo?) was going to be my first suggestion. Are you absolutely 100% definitely sure about that?

I can't imagine what that could be. Any chance you could post a video?

Something like this:

The PodX is what you need. The ES-9 is too deep for the Pod.

It's busy with A and preloading the start of B.

Quite likely yes.

In the meantime, have you tried running at a higher sample rate?

I would urge you towards the disting EX which achieves 700µs latency.

No. DIN sync uses 2 of the 8 output channels - there's no way around that.

Yes it will work. See viewtopic.php?f=35&t=90403

Are those 48kHz samples we're talking about? The latency seems in the right ballpark. It will depend on card speed to some extent. Are you using the recommended playlist settings from the manual? -loop=0 -gap=0 -fadeIn=0 -fadeOut=0 -retriggerOnSampleChange=0 Also, are we talking about a disting mk4 ...

Would it be worth trying macOS 10.15.6?

You don't need to ground the shield of a TRS MIDI plug, nor do systems need to share ground. That's the joy of MIDI - it's opto-isolated.

Add the Effect version after the Instrument version.

It would also help to see the ES-9 config, and the Aggregate setup in the Audio MIDI Setup utility.

A batch just left the factory. They should find their way out to stores very soon.

You would direct the MIDI directly to the individual plug-ins, not to the track input.

Why would you need to channelize the messages?

What happens if you use it in standalone mode (without the mac)? E.g. just route some audio through it to your speakers. Does it cut out after a while? If so, the issue is likely not to do with the USB connection/computer.

The difference with some of those (e.g. GB_Hapi_vs_Xylophone) is that the samples are 24 bit, which is putting 50% more pressure on the card compared to 16 bit samples. Though FWIW that set is stable here up to 7 voices - it breaks up like your recording at 8 voices. That's using a Kingston Canvas S...

You need to raise the 'Gate offset' parameter.

How much latency are we looking at, in ms?