Anyone having any latency issues with USB?

Jono Bacon

Inspired
Hi Everyone,

I am having an unusual issue. I am currently testing my Axe FX III plugged into a MacBook Pro and have the buffer setting at 256 to get as low latency as possible. The current latency figures are:

Input: 8.583
Output: 7.354
ASIO Guard: 0.00

These figures seem pretty good to me.

Strangely, it seems like my recorded audio is slightly...off. Now, I have tested this in multiple ways:

* Recording the same piece of music at different tempos to ensure it is locked in.
* Recording one part with drums, muting it, and then recording the double part.
* Recording stabs of audio on the beat.

I am noticing that my recordings sound sloppy and without blowing my own trumpet, I am a tight rhythm player.

Of course, I want to discount my own playing - any recommendations for how I can effectively test this?

Is anyone noticing anything like this?
 
Hi Everyone,

I am having an unusual issue. I am currently testing my Axe FX III plugged into a MacBook Pro and have the buffer setting at 256 to get as low latency as possible. The current latency figures are:

Input: 8.583
Output: 7.354
ASIO Guard: 0.00

These figures seem pretty good to me.

Strangely, it seems like my recorded audio is slightly...off. Now, I have tested this in multiple ways:

* Recording the same piece of music at different tempos to ensure it is locked in.
* Recording one part with drums, muting it, and then recording the double part.
* Recording stabs of audio on the beat.

I am noticing that my recordings sound sloppy and without blowing my own trumpet, I am a tight rhythm player.

Of course, I want to discount my own playing - any recommendations for how I can effectively test this?

Is anyone noticing anything like this?
Are you recording to re-amp? I noticed when I was reamping that one of the tracks (I think it was the direct track for re-amping) had some latency. Could that be it?
 
Hi Everyone,

I am having an unusual issue. I am currently testing my Axe FX III plugged into a MacBook Pro and have the buffer setting at 256 to get as low latency as possible. The current latency figures are:

Input: 8.583
Output: 7.354
ASIO Guard: 0.00

These figures seem pretty good to me.

Strangely, it seems like my recorded audio is slightly...off. Now, I have tested this in multiple ways:

* Recording the same piece of music at different tempos to ensure it is locked in.
* Recording one part with drums, muting it, and then recording the double part.
* Recording stabs of audio on the beat.

I am noticing that my recordings sound sloppy and without blowing my own trumpet, I am a tight rhythm player.

Of course, I want to discount my own playing - any recommendations for how I can effectively test this?

Is anyone noticing anything like this?

rout the metronome/click of axe fx to a channel in cubase. sync the axe fx to cubase's tempo. if the clicks are on time / on beat this would mean axe fx is reporting the latency accurately to cubase and cubase is accurately compensating the input latency.
 
rout the metronome/click of axe fx to a channel in cubase. sync the axe fx to cubase's tempo. if the clicks are on time / on beat this would mean axe fx is reporting the latency accurately to cubase and cubase is accurately compensating the input latency.

I thought about this, but I don't think the Axe FX III has a metronome or click, right?

I guess I could feed an external click into the Axe FX maybe?
 
Try playing any track and cabling output 1 to input 1. Don't place an input 1 block in the preset for this test. Record a new track with USB channel 5 as input (unprocessed input 1). Is the new track significantly later than the original? Is it consistent throughout several tests?

If it's consistent you should be able to set the latency compensation so it's not an issue.
 
i am currently testing my Axe FX III plugged into a MacBook Pro and have the buffer setting at 256 to get as low latency as possible.

The USB buffer size on the Axe-Fx III itself goes all the way down to 8. 256 is the highest setting and is the (conservative) default value.
 
The USB buffer size on the Axe-Fx III itself goes all the way down to 8. 256 is the highest setting and is the (conservative) default value.


latency or setting of axe fx should't yield into any printing timing issues in DAW though. Any modern DAW is compensating for it, right?
 
I thought about this, but I don't think the Axe FX III has a metronome or click, right?

I guess I could feed an external click into the Axe FX maybe?

oh sorry, didn't mess much with that side of Axe 3 yet, so i didn't know there is no metronome.

But yes, use an external click. That would be the best way to test and see whats going on.

Change the buffer size, regardless of it's setting the printed click of metronom should be on same position on cubase ruler. if not it means the compensation isn't working and yes than it messes with your timing-playing-feel.

I would try to help and do the same test but i am out of town for some days.
 
Last edited:
latency or setting of axe fx should't yield into any printing timing issues in DAW though. Any modern DAW is compensating for it, right?

DAWs only compensate for the latency reported by the audio driver. This is additional external latency on the Axe-Fx III itself, dictated by Setup->I/O->USB Buffer Size (well and the DAC/ADC)

The USB Audio Class 2.0 spec does have a way for a unit to report latency, but as far as I can see this is not supported by the current USB descriptors of USB firmware 1.06 (bmControls is 0x0 for both INPUT and OUTPUT terminal descriptors).
If supported, a TE_LATENCY_CONTROL query would report the latency for the input and output terminals, which should include ADC/DAC latency as well as the USB Buffer size.

In any case, most DAWs however have manual offsets to compensate for additional external latency. You can use that to compensate for USB Buffer size. In Reaper for example, Options->Preferences->Audio->Recording has "Output Manual Offset", "Input Manual Offset".

It seems the Axe-Fx III actually uses 2 buffers of size "USB Buffer Size". Or at least that seems the case with an experiment I ran with Reaper. On Windows 10, using a ASIO Buffer size of 64, Safe mode disabled. I disabled "Use audio driver reported latency" in Reaper. Then I set USB Buffer Size to 256 at the Axe-FX III.

I then set up a patch with IN USB Block going to OUT2 block. In Reaper I added one track with a send to USB 7+8 (i.e. IN USB) and another track set to record IN 3+4 (i.e. OUT2 Block).

After recording I measured the latency in Reaper, which was about 1238 samples. The FASUSBAudio Control panel reported as 136 samples for input, 72 samples for output for latency, accounting for 208 samples. That leaves around 1030 samples, which could be explained if we assume there are 4 USB Buffers at the Axe-Fx III (2 for input, 2 for output), i.e. 1024 samples.

I dropped the USB Buffer size to 128 and 64, where I measured 728 and 470 samples of latency respectively, which corroborate the idea.

All that basically to say... in your DAW, to account for this external latency, use 2*USB Buffer Size samples + around 40-50 samples for DAC/ADC for Input and output latency offsets.
 
Last edited:
All that basically to say... in your DAW, to account for this external latency, use 2*USB Buffer Size samples + around 40-50 samples for DAC/ADC for Input and output latency offsets.
Hm... how's this math work? 2 * 256 + 50 = 562... but it's supposed to be 1238, no?
Am I doing something stupid here? :D
 
Hm... how's this math work? 2 * 256 + 50 = 562... but it's supposed to be 1238, no?
Am I doing something stupid here? :D

562 for input latency offset, 562 for output latency offset.

I measured 1238 for the loopback, so it’s combined input and output latency. And since I stayed USB only it doesn’t include ADC or DAC latency.
 
Hi Everyone,

I am having an unusual issue. I am currently testing my Axe FX III plugged into a MacBook Pro and have the buffer setting at 256 to get as low latency as possible. The current latency figures are:

Input: 8.583
Output: 7.354
ASIO Guard: 0.00

These figures seem pretty good to me.

Strangely, it seems like my recorded audio is slightly...off. Now, I have tested this in multiple ways:

* Recording the same piece of music at different tempos to ensure it is locked in.
* Recording one part with drums, muting it, and then recording the double part.
* Recording stabs of audio on the beat.

I am noticing that my recordings sound sloppy and without blowing my own trumpet, I am a tight rhythm player.

Of course, I want to discount my own playing - any recommendations for how I can effectively test this?

Is anyone noticing anything like this?
On DAWs, if you have input monitoring on you will get some latency, even at 256. I never record with input monitoring on. Thunderbolt is replacing USB in audio interfaces but a computer motherboard has to support it. Axe 3 is USB still because most don't have Thunderbolt supported Motherboards on their computers. Turn input monitoring off on your DAW track and that should solve your problem.
 
On DAWs, if you have input monitoring on you will get some latency, even at 256. I never record with input monitoring on. Thunderbolt is replacing USB in audio interfaces but a computer motherboard has to support it. Axe 3 is USB still because most don't have Thunderbolt supported Motherboards on their computers. Turn input monitoring off on your DAW track and that should solve your problem.

this is about recorded audio, not about monitoring while playing.
 
All that basically to say... in your DAW, to account for this external latency, use 2*USB Buffer Size samples + around 40-50 samples for DAC/ADC for Input and output latency offsets.

Thanks, @AlbertA. Shouldn't the driver correctly report the latency so the DAW can automatically offset?

Are you saying the axe driver doesn't report this correctly and therefore we need to set this in our DAWs?
 
Thanks, @AlbertA. Shouldn't the driver correctly report the latency so the DAW can automatically offset?

Indeed. Ideally the driver should report all of the latencies involved. Not many interfaces get this right however.

Are you saying the axe driver doesn't report this correctly and therefore we need to set this in our DAWs?

Right, It doesn't report all of the latency.
For example, in windows, it only reports the ASIO buffer size.
I assume it's similar with Apple's driver in OSX, where it may just report Core audio's buffer size.

Now, it appears the USB 2.0 UAC2 spec does provide a way for a UAC2 device to report its latency, but it's not clear if implemented that would be supported by Apple's driver. In Linux, ALSA doesn't seem to query this at all, so someone would have to add support for it. In Windows, FAS controls the driver so they could potentially make it work.

This becomes less of an issue as you lower the USB Buffer Size in the Axe-Fx III however.

But the best you can do (with any audio interface) is to record the loopback and measure the latency yourself.

You can easily do this with the Axe-Fx III by connecting a cable from say output 3 L to input 2 L and create a preset that routes IN USB->OUT 3 Block. Make sure to increase the OUT 3 knob in the front.

Then in your DAW, set a click track or something with clearly defined transients, and send it to track 7 (IN USB L).
In another track, set it to record Input 7 (INPUT 2 L).

In Cubase you should adjust "Record Shift" by the amount of delay between the tracks. Re-Record the track, check and adjust as necessary until both tracks are aligned.
 
Last edited:
Does the amount of effects, amp sims etc effect the processing delay?


I guess I can test it when attempting to adjust Logic.
 
Back
Top Bottom