Go Back   Cockos Incorporated Forums > REAPER Forums > Recording Technologies and Techniques

Reply
 
Thread Tools Display Modes
Old 10-25-2019, 08:17 PM   #1
Reason
Human being with feelings
 
Join Date: Apr 2015
Posts: 976
Default AXE IO - High Latency, 261ms - is it my methods (SOLVED, it was my methods)?

Hardware:
AXE IO connected via USB to a USB port on my motherboard (ie, not an addon card)
I7-6700
16GB RAM
Win10 with latest updates

Connection:
1. Guitar cable (not TRS) from Input 1 to Output 1 (Input is line level, not the low-Z)
2. XLR mic from Input 2 to Output 1

Method 1:
Track 1: Click source
Track 2: Recording Input 1 (line in)
Track 3: Redo of Track 2 with my old audio card's software exited.
Track 4: Recording Input 2 (mic) (held on my head to emulate air gap between speakers and ears)
- These two were recorded in separate takes, not simultaneously

Zoomed all the way in and selected from the very start of the first audio on Track 2 or 3 to the very start of the first click source, set Transport to samples.

Results: 11741 samples for the mic, 12044 samples for the DI (and why is THAT backwards?) which is about a 1/4 second, or almost an 8th note at 120 bpm.

This is visible in the wave forms, too.

This user from August is getting a 10th of what I'm getting:
https://www.gearslutz.com/board/show...postcount=3413

Playing 1/4 notes with my guitar does line up with the recorded click. Just playing without recording is very responsive, no latency that I can detect.

Haven't run the tool RTL tool from Oblique Audio yet. Is a TRS cable required, or is a regular TS guitar cable fine?

Thoughts/suggestions/comments? Thanks!
Attached Images
File Type: jpg Latency.JPG (49.0 KB, 289 views)

Last edited by Reason; 11-03-2019 at 03:08 PM.
Reason is offline   Reply With Quote
Old 10-26-2019, 02:20 AM   #2
Stella645
Human being with feelings
 
Stella645's Avatar
 
Join Date: Sep 2008
Location: UK
Posts: 3,648
Default

Quote:
Originally Posted by Reason View Post
Hardware:

Playing 1/4 notes with my guitar does line up with the recorded click. Just playing without recording is very responsive, no latency that I can detect.
Which tells us that your test results and/or conclusions are clearly incorrect.
Stella645 is offline   Reply With Quote
Old 10-26-2019, 10:16 AM   #3
Reason
Human being with feelings
 
Join Date: Apr 2015
Posts: 976
Default

Quote:
Originally Posted by Stella645 View Post
Which tells us that your test results and/or conclusions are clearly incorrect.
OK, that's what I was thinking, but I am not sure why. Any ideas?
Reason is offline   Reply With Quote
Old 10-26-2019, 11:45 AM   #4
Stella645
Human being with feelings
 
Stella645's Avatar
 
Join Date: Sep 2008
Location: UK
Posts: 3,648
Default

Remove all FX from project (and monitoring FX) and try again with line out of interface to line in. Use an audio click on track 1.

Even when done correctly your results don't show latency as Reaper is already compensating for latency....any difference is just between reported (which you can see top right of arrange window) and measured latency.
Stella645 is offline   Reply With Quote
Old 10-26-2019, 04:21 PM   #5
Reason
Human being with feelings
 
Join Date: Apr 2015
Posts: 976
Default

Thanks. There are no effects in the project, except for the guitar track, but that was added after the other tracks.

I'm using this method:
https://wiki.cockos.com/wiki/index.p...erface_Latency

I may not have stated it clearly, but I'm running the line out of the interface to the line in and recording the result on Track 2. Track 3 is the same, it was just done after I uninstalled my previous card's driver. Track 4 is recording the speakers with a mic connected to In 2, and I'd expect that to have slightly more latency than tracks 2 or 3 due to the extra 3 feet the sounds has to go through the air, but it's actually a few hundred samples less.

Quote:
Even when done correctly your results don't show latency as Reaper is already compensating for latency....any difference is just between reported (which you can see top right of arrange window) and measured latency.
I don't quite understand, can you elaborate on this? It seems likely that not (or mis)understanding this is the root of my "problem."
Reason is offline   Reply With Quote
Old 10-27-2019, 05:29 AM   #6
Stella645
Human being with feelings
 
Stella645's Avatar
 
Join Date: Sep 2008
Location: UK
Posts: 3,648
Default

And you're 100% sure no plugs in Monitoring FX??

Delete everything except track 2. Cut and move the audio click manually to the grid.

Turn back on use driver reported latency.

Now record track 1 to a new track and compare. How's that look??
Stella645 is offline   Reply With Quote
Old 10-28-2019, 01:19 PM   #7
Reason
Human being with feelings
 
Join Date: Apr 2015
Posts: 976
Default

Thanks, turns out I had left ReaInsert on track 1 from when I was trying to measure latency that way. I took it off and measured 34 samples difference between click source and recorded audio. Haven't had time to run more tests but looks like that was my issue.

ReaInsert showed about 130 ish samples difference, so I'd like to do some more testing to see if I can narrow down what I should put in as compensation, but a difference of 100 samples works out to 2ms at 48khz, so not mission critical.

Thanks for your help!
Reason is offline   Reply With Quote
Old 11-03-2019, 03:28 PM   #8
Reason
Human being with feelings
 
Join Date: Apr 2015
Posts: 976
Default

Finally had time to go back and do some more testing, and I'm following up for any googlers out there.

Settings: 48Khz, 32 block size

Using the ReaInsert method, 7 minute video here:
https://www.youtube.com/watch?v=TzS--D765Zw
ReaInsert says the device reported latency is 424 samples. Ping shows an addtional 166 samples.

Using the "manual" method here:
https://wiki.cockos.com/wiki/index.p...erface_Latency

and holding the mic on top of my head to simulate what I'd hear coming from my speakers, I measured 276 samples difference between the generated click source and the recorded click source.

I believe the difference of 110 samples corresponds to the sound traveling through air over the distance between my ears and the speakers, which is little less than a meter.

Because I sometimes track with headphones and sometimes with speakers, I decided to use a value between the two values, and added 200 samples to the Input latency in Preferences | Audio | Recording | Input Manual Offset.

Unless I'm wrong, which I very well might be, this calculates to an RTL of 13ms at 32b/48khz, which isn't great, and this is before any plugin-related latency. Still, I think it's workable.
Reason is offline   Reply With Quote
Old 11-03-2019, 10:58 PM   #9
zeekat
Human being with feelings
 
zeekat's Avatar
 
Join Date: May 2009
Location: Polandia
Posts: 3,578
Default

Quote:
Originally Posted by Reason View Post
Unless I'm wrong, which I very well might be, this calculates to an RTL of 13ms at 32b/48khz, which isn't great, and this is before any plugin-related latency. Still, I think it's workable.
13ms at 32 buffer is a bit lame, especially if it's marketed for guitarists who'd like their monitoring nice and fast. That'd be worse than my 2011 and not-guitar-specific Mackie, you sure your measurement is good again?
__________________
AM bient, rund funk and heavy meteo
my bandcamp+youtubings
zeekat is offline   Reply With Quote
Old 11-04-2019, 08:51 AM   #10
Reason
Human being with feelings
 
Join Date: Apr 2015
Posts: 976
Default

Not positive, but not sure what else to do. However, while I don't fully trust my memory, I had thought that ReaInsert showed 240 samples reported latency on my first round of testing, and 106 samples on the ping. I'm positive the ping was different, but not sure about the reported latency.

One thing I'm still not sure about is whether or not using a TRS cable matters.
Reason is offline   Reply With Quote
Old 11-04-2019, 10:22 AM   #11
Stella645
Human being with feelings
 
Stella645's Avatar
 
Join Date: Sep 2008
Location: UK
Posts: 3,648
Default

TRS would either work or not work for this purpose. If you're record a click and it sounds like the same click then its working.

I think something may be wrong somewhere as 13ms at 32 buffer does seem too high.
Measure with RTL Utility to get a baseline and work from there.
Stella645 is offline   Reply With Quote
Old 11-05-2019, 07:40 PM   #12
Reason
Human being with feelings
 
Join Date: Apr 2015
Posts: 976
Default

OK, I think I got this sorted.

RTL Utility shows about 7.3 ms at 48khz / 64, which is fine.
Code:
"Type","Device","S.R.","Bits","Buffer","Rep. Input Lat.","Rep. Output Lat.","Rep. RTL","Measured RTL","Measured RTL (ms)","Estimated RTL","Return Loss","Noise Floor","Valid","Date/Time","Num. Inps","Num. Outs","Num. I/O","Out Ch","Inp Ch","Driver","Comment","OS Factor"
"ASIO","AXE IO ASIO",48000,32,64,136,120,256,353.4,7.363,353,22.8,-92.0,1,5 Nov 2019 18:37:01,1,1,2,"-not implemented-","In 1","-not implemented-","-not implemented-",16
"ASIO","AXE IO ASIO",48000,32,64,136,120,256,348.6,7.263,353,22.8,-110.3,1,5 Nov 2019 18:37:12,1,1,2,"-not implemented-","In 1","-not implemented-","-not implemented-",16
"ASIO","AXE IO ASIO",48000,32,64,136,120,256,348.6,7.263,353,22.8,-111.1,1,5 Nov 2019 18:37:14,1,1,2,"-not implemented-","In 1","-not implemented-","-not implemented-",16

Quote:
I had thought that ReaInsert showed 240 samples reported latency on my first round of testing, and 106 samples on the ping.
I realized that the reported ping displayed by ReaInsert includes any offset added under Prefs | Recording | Input Manual Offset. So that's why it was different.
Reason is offline   Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -7. The time now is 02:15 AM.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.