Latency is one of those terms we hear all the time with regard to many different aspects of todays DAW and our digital audio recording workflows. We all know what we think latency is but how do companies come up with their latency figures be they good or bad? Take a look at the chart below which we found on the Clarett page of the Focusrite web site.
In these tests Pro Tools performed significantly better than the other DAW applications so we asked Focusrite if they would mind letting us know what the tests and procedures were that they went through to get these numbers. Will Evans from Focusrite came back to us.
When we did all of the latency measurements for the Claretts, the DAW was the only thing that changed in the test setup. All other test conditions were fixed. Any variance between round trip latencies has to be down to the different DAW layer and is out of our control. Only the DAW manufacturers know exactly how their DAWs work and whether or not there is any internal buffering taking place in addition to the sample buffer set by the user, depending on how the DAW has been designed.
In terms of how we measure our round trip latency. Let me explain. We use two audio interfaces (A & B) both connected to a DAW. Interface A is the interface we want to measure the RTL of so the DAW on interface B is fixed and used to take the measurements. Interface A is connected to the system we want to measure so this is where the DAW varies. We set up the cabling as follows:
Connect device B's output 1 to A's input 1
Connect device A's output 1 to B's input 1
Connect device B's output 2 to B's input 2 in order to act as a control signal to compare against
Then we send a single sine wave oscillation from device B. We record 2 signals back in.
1. The signal from device B output to input, and
2. The signal from device B output that goes into A, through the DAW on the system we are measuring, back out of device A’s outputs and back into device B’s inputs.
The result is that the test signal appears at slightly different times on each input channel. The difference between the two is the total round trip latency of device A.
So there you have it. No smoke and mirrors just a real life test. As we always say when testing or fault finding, the only true way to test a system is to change just one thing at a time as Focusrite have done in their testing.
For more information on latency basics then read this excellent article by community member David Finnamore