How Do I Perform a Loop Back Test in Reaper?

[SEA]

Trying to perform my first converter loop back test in Reaper using my Steinberg UR28M.

Any tips?

Thanks!

SEA

[xpander]

Check this thread by Fabian:
Loop-back testing JS FX
http://forum.cockos.com/showthread.php?t=63318

If you use the project included in the package, it may be as simple as connecting the left output of your interface to the right input and hitting Play. I tested it and it works here without any extra settings but better check yours if not sure.

[serr]

Uncheck 'Use audio driver reported latency' in Reaper Preferences/Audio/Recording and make sure there is no manual offset entered either.
This feature is intended to be used to align newly recorded audio AFTER YOU KNOW YOUR SYSTEM LATENCY.

Put up a sample in a track with an obvious easy to see transient peak when you zoom in to the sample level.

Patch your output to an input. (Mind your levels.)

Record the sample to a new track from said input.
(Original track with sample > output > patched to input > input assigned to new track)

Now zoom in. You'll see the new recorded item later in the timeline vs. the original. Mark the original peak in track 1. Mark the new recorded sample peak in track 2. Double click in the timeline to make a time selection between markers. Look at the time selection length (in samples) reported in the transport.

Do this at every sample rate to see what your system can do and where the 'line in the sand' is for minimum required block size for stability.

You will have a different latency for each sample rate as expected.

For live sound work, you need your total system latency to be 11ms or less.
Get a baseline for the system itself with no plugins. If your hardware cannot stably run at 11ms or less with no plugins - then you need to buy some upgrades!

Add the plugins you aim to use next. If one or more plugins require a larger block size to operate with stability - this may mean certain plugins cannot be used for live sound.

For studio post work, latency is a moot point entirely but you still need your recorded audio to line up with what's already recorded on the screen. This is what that offset feature is for in Reaper Preferences/Audio/Recording.

Tip:
If (after determining the latency) you check the box for 'Use audio driver reported latency' in Reaper Preferences/Audio/Recording, you will find that it will automatically 'normalize' alignment between the different sample rates and the result will be off by the same amount (the guess the driver makes - 'guess' because it can't know what hardware is connected over digital audio connections) for every sample rate. You can now manually enter the offset (in addition to the 'Use audio driver reported latency') and you should not need to change the offset manually as you use different sample rates.

If you don't use 'Use audio driver reported latency', you will need to enter the unique offest for each sample rate every time you switch.


Here's what my notes look like for my loopback tests for reference:

(I noted the latency Reaper reports at the top of the screen for reference. This of course is the latency Reaper adds and does not include system hardware as Reaper has no way of getting feedback for such a thing. You will often see forum posts where someone assumes this is actually the system total and then you get the claims that 11ms lag is perceptible. What's actually going on is the total lag in that scenario is much higher than 11ms.)

Apogee Rosetta 800 192k

96k 1024 samples = 2730 samples (28ms) Reaper = 12/12ms
= 362 samples with 'Use audio driver reported latency'
96k 128 samples = 938 samples (9ms) Reaper = 3.3/2.6ms
= 362 samples with 'Use audio driver reported latency'

88.2k 1024 samples = 2666 samples (30ms) Reaper = 13/12ms
= 325 samples with 'Use audio driver reported latency'
88.2k 128 samples = 875 samples (9ms) Reaper = 3.4/2.7ms
= 325 samples with 'Use audio driver reported latency'

48k 1024 samples = 2400 samples (50ms) Reaper = 23/23ms
= 160 samples with 'Use audio driver reported latency'
48k 128 samples = 608 samples (12ms) Reaper = 3.3/2.6ms
= 160 samples with 'Use audio driver reported latency'

44.1k 1024 samples = 2374 samples (53ms) Reaper = 25/25ms
= 150 samples with 'Use audio driver reported latency'
44.1k 128 samples = 582 samples (13ms) Reaper = 3.3/2.6ms
= 150 samples with 'Use audio driver reported latency'



MOTU 828mk3 (1x or 2x aggregate)

analog inputs
96k 1024 samples = 2215 samples (23ms) Reaper = 10/10ms
= 125 samples with 'Use audio driver reported latency'
96k 128 samples = 423 samples (4ms) Reaper = 1.5/1.5ms
= 125 samples with 'Use audio driver reported latency'

Apogee AD-16 -> ADAT inputs
96k 1024 samples = 2257 samples (23ms) Reaper = 10/10ms
= 167 samples with 'Use audio driver reported latency'
96k 128 samples = 465 samples (4ms) Reaper = 1.5/1.5ms
= 167 samples with 'Use audio driver reported latency'

88.2k 1024 samples = 2257 samples (25ms) Reaper = 11/11ms
= 167 samples with 'Use audio driver reported latency'
88.2k 128 samples = 463 samples (5ms) Reaper = 3.4/2.7ms
= 167 samples with 'Use audio driver reported latency'

48k 1024 samples = 2198 samples (46ms) Reaper = 21/21ms
= 115 samples with 'Use audio driver reported latency'
48k 128 samples = 406 samples (8ms) Reaper = 3.0/3.0ms
= 115 samples with 'Use audio driver reported latency'

44.1k 1024 samples = 2199 samples (49ms) Reaper = 23/23ms
= 116 samples with 'Use audio driver reported latency'
44.1k 128 samples = 407 samples (9ms) Reaper = 3.8/3.7ms
= 116 samples with 'Use audio driver reported latency'


48k 192 samples = 534 samples (11ms) Reaper = 4.3/4.3ms
= 115 samples with 'Use audio driver reported latency'



Apogee Rosetta 800 + MOTU 828mk3 + MOTU 828mk3 #2

48k 192 samples = 743 samples (15ms) Reaper = 6/6ms
= 115 samples with 'Use audio driver reported latency'

96k = doesn't work at any buffer size

Also…
2048 sample buffer size with the Apogee 800 results in 272 sample dropouts every 2048 samples.

[DarkStar]

^^^^

I thought that I'd give this a try. My Delta 66 has a Monitor output which can be used to send the audio back into Reaper. I selected that input on a second Reaper track and dragged the level right down (to avoid any feedback loop).

Here's the result:
>>> https://i.imgur.com/1xbCLNf.png

The top track is the original clip and the lower one is the recording. The recording is earlier than the original, as you can see in the zoomed-in section on the right-hand side (by 71 samples).. Have I gone back in time?

Mu buffer size is 256 samples, 6.7/6.4ms ASIO.

[serr]

Did you remember to stop Reaper moving it by unchecking 'Use audio driver reported latency'?

[DarkStar]

D'oh! Missed that (even though I read it in your post!)

I tried again and get a lag of 516 samples

[Fabian]

Originally Posted by xpander

Check this thread by Fabian:
Loop-back testing JS FX
http://forum.cockos.com/showthread.php?t=63318

If you use the project included in the package, it may be as simple as connecting the left output of your interface to the right input and hitting Play. I tested it and it works here without any extra settings but better check yours if not sure.

It should work that simple
But if you have any issues or questions, just ask.

[gmgmgm]

I found this a quick and simple solution for just testing interface latency:

http://www.oblique-audio.com/free/rtlutility

From Oblique Audio's description:

"RTL Utility is a tool for measuring the Round Trip Latency of your DAW and audio interface. The utility has been developed for the Low Latency Performance test at dawbench.com."

Though it seems to me that it removes the DAW from the test process.