How to test actually latency (tracking, monitoring through PT)?

[smpkeys]

I've been reading threads all day trying to figure out how to determine the actual latency that my audio interface (s) are getting. (not just what the buffer say's).

I found a tool on the windows side, but nothing for mac:

http://www.centrance.com/products/ltu/

I've done lots of reading over at dawbench.com too, and found many charts and users reporting figures like "they are getting 1.5 ms at a 32 buffer, 2.88 at 64, etc.." - But I can't figure out how they find that. I've tried the loopback test at several buffer settings, but i get the same result (57 samples) at every buffer, so I'm not sure that's the correct way to find what I'm looking for.

I'm trying to determine what the 2 interfaces I have are getting on my system, so I know what to compare too as I try and find the best interface solution for me.

thanks!

[Shan]

All versions of PT will compensate for the input delay induced by the audio interface A/D conversion after a record pass. You'll have to bypass this by using an Aux in your loop back test. Follow this, but route back into an Aux track instead. Buss the Aux to an Audio track and record on that track. This will bypass the record compensation. Sessions at different sample rates will also yield different results. Here is a "general" latency chart by Avid. Take it with a grain of salt and do your own loop back benchmarks at different session sample rates with your hardware I/O.

Shane

[smpkeys]

Thanks so much Shan! That's exactly what I needed.

I'm at 48 k on my test session, so I think to get the RT latency, I divide the number of samples of delay by 48, right?

[smpkeys]

bump

[Shan]

Do a loop back test of a printed click track and measure the result in samples in the Edit Window. You'll have the real latency doing that.

Shane

[chrismeraz]

Originally Posted by Shan

Do a loop back test of a printed click track and measure the result in samples in the Edit Window. You'll have the real latency doing that.

Shane

I did exactly that, but I'm reading 44ms. It's about ten times higher than I expected!

I'm using an Apollo Firewire on a Windows 10 machine with 64 samples of buffer at 96kHz. I was expecting latency around 5ms...

[chrismeraz]

Here's what my screen looks like with a loopback at the different sample rates and buffers. Weird, right?

[YYR123]

I need to try this....and see what happens.

I wish i had chartered all of my interfaces up till now

[chrismeraz]

I found the problem. I had a UAD effect (EMT 140 Reverb) inserted elsewhere in this session. When I removed it, latency fell to around 4ms.

In other words, if I use any UAD plugins in Pro Tools, latency increases drastically. I can use only native plugins if I want to keep latency down. Is this normal behaviour?

Here's a screenshot.

• First line is the recorded click
• Second line is the loopback latency at 96k and 64 samples of buffer
• Third line is the loopback latency after adding any UAD plugin anywhere in the session

You can see that the latency went from 4ms to 48ms.