Let's talk about latency

Hi all,
I’d like to know what you think about latency when we use virtual instruments.
On my MacBook Pro 14’ M1Max, I use Kontakt with gig Performer and the Nord piano bank from Junior Porciuncula.
With my Fireface 802, I have a latency of 2.5 ms (44.1 kHz - 64 samples) and the playability is top notch with no crackling. The processor never exceeds 13%.
For practical reasons, I invested in a Key-Largo over a year ago and I’ve been able to do all my concerts with a latency of 4ms (96Khz - 64 samples). In this particular case, the processor is a little more stressed but never exceeds 20-21%.
I’ve just invested in a Yamaha CP88 with a sound card (only at 44.1 Khz), and I’m also running at 4ms at 64 samples.
I have the impression that I don’t feel any difference when I play with the CP88 card (4ms) or my Fireface 802 (2.5ms) but I still think it’s better with the Fireface.
I’m quite tempted by the idea of selling my Key-Largo because the CP88 does the job perfectly (as keyboard, master keyboard and sound card).

I’ve got now an important question because sometimes I feel like I’m psyching myself out :-/, what do you think is the threshold you shouldn’t exceed to avoid losing playability ?

I’d like to take this opportunity to give some output references for the 4 cards I tested (info returned by Kontakt):
CP88 (output 2.5ms - 44.1Khz)
Fantom-06 (output 2.2ms - 44.1 / 1.9ms 48Khz / 2.2 - 96 Khz)
Key-Largo (output 3.1ms - 44.1 / 3.0ms - 48 Khz / 3.0ms - 96 Khz)
Fireface 802 (output 1.0ms - 44.1 / 0.9ms - 48 Khz / 0.8ms - 96 Khz)

What does Gig Performer’s Latency Tool say?

Same result than Kontakt. :wink:
CP88 => 4ms : ouput 2.5ms + 1.5 treatment (64 samples)
Fireface 802 => 2.5ms : ouput 1.0 ms + 1.5 treatment (64 samples)

Gig Performer | Audio latency, buffer size and sample rate explained

Don’t go too far away from the speakers if 2.5ms is a problem for you! :wink:


Guitarist here. I only begin to notice it around 10ms in GP, but that still feels more or less like having an amp behind me or across the stage from me. When it gets upwards of 15-20ms, I start to feel really uncomfortable, and above that it becomes hard to deal with. That’s for doing fairly rapid, precision picking, if I’m just rocking out I can handle a bit more. And live with a band is more forgiving than studio, or solo performance.

There is other latency, from your interface, and from the distance to your speakers, and maybe other factors (including plugins, of course there’s a good thread on zero-latency plugins). So what you’re seeing in the GP stats isn’t the only latency to consider.


I play with ear-monitor, think it’s close enough to my ears :wink: :smile:
Okay, I’ll stop playing around with this.

Welcome to the club! I’m a latency nerd, too:

In short: A bit of a constant latency is like some distance to the sound source and isn’t the main problem. Otherwise, no one could play Bach Toccata and Fugae on a church organ, ever.

The problem would be latency jitter, which is like a random timing or unsteady groove. This jitter could happen by improper handling of the audio blocks and incomming MIDI in the software, and this would go really worse with larger buffer settings. This was the case with the discontinued former VST host Forte, which forced my hunt for lowest possible buffer size, in the past.

As you found out, too, RME has exceptional low-latency interfaces and drivers. Some others too, just to say. And GP has a very low latency jitter MIDI to audio engine, so as long you use a resonable low buffer size, latency isn’t an issue.

1 Like