« A responsive keyboard: getting 2 millisecond latency | Main | Global Warming - Where's the heat? »

Mar 29, 2015

Comments

Ken Rushton

The latency "felt" consistent in both cases, once I got used to it, although for what it's worth faster felt a lot better.

The DPC latency checker is really useful for checking for jitter: I gather it checks how fast the operating system responds.

My little Acer netbook on battery power has an horrible intermittent jitter - caused by a bug in the Microsoft-supplied battery-checking driver that activated when on battery. This was clearly shown by DPC latency checker.

I infer that MS never checked the latency of their battery-test drivers while actually on battery power. Having been a tester, I can see how that might be overlooked; it's a fair bit of fiddle to setup tests on machines on battery. I kept hoping it would be fixed, to no avail.

And yes, the latency was consistent for every note. I did it dozens of times, always seeing the same gap.

I'm glad I got a good response. this "little" posting took about a day to do.

Ken.

BrettThePark

For *most* people 20-30ms is imperceptible. You might actually want to test the delay amount multiple times. Often the audio generation pipeline will also have something called jitter. Jitter is basically the inconsistency of producing an audio signal. For example, if you do this test 10 times with a single note, you might get a range of delays from 20-45 ms giving you a 25ms jitter. The jitter is often what people notice as your brain will automatically correct for a consistent delay after a bit of time, but it will not correct your perception for jitter.

The comments to this entry are closed.