iPad recording latency… Help me

[metropolis_4]

I used to use GarageBand on my old iPad to record and it worked great. With my new iPad I can’t figure out how to make it work.

Using my fm3 as an AI I’ve got instrument plugged into the fm3 -> usb to iPad -> monitoring from the output on the fm3. The problem is I’m getting latency somewhere so when I play back something I recorded it is behind the beat.

I’m trying to figure out why this same thing worked fine with my old Lightning iPad but it doesn’t work with the new usb-c one. Is there some difference in how the two work?

 

[paisleywookiee]

Crap, I can’t try it right now, but I will first thing tomorrow if someone doesn’t answer you. FWIW, I’ve got the FM9 and an iPad A16 (current base model with USB-C).

 

[bikescene]

The USB recording function of the FM3 doesn’t report the latency of the blocks it seems. It reports the device latency, but I think Amp+Cab blocks add something in the 40ms range. If you search around on the Fractal forums, you’ll find threads on the topic.

I think it got fixed on the FM9 and Axe-FX III, but is still an issue on the FM3. I unfortunately couldn’t find a way to enter latency offset in GarageBand to automatically compensate for this problem.

What worked for me when recording and overdubbing with the FM3 as an audio interface on a computer or iOS device:
-Load up your typical FM3 patch with an amp and cab block. Pick a clean amp.
-Pick a test audio file with an identifiable transient like an IR.
-Send the audio file to Output 3+4 on the FM3.
-With the Input 1+2 Source in the I/O setting set to USB, record arm and record the reamped signal.
-Figure out the offset required to align your recorded signal with the test audio file.

On a full fledged DAW, I can figure out the exact number of samples to enter an offset into the DAW settings (somewhere in the 2100 range). On my iOS recording app (Cubasis), adjusting for 49ms gets it close enough for me. I think I typically have UltraRes IRs pulled up when checking this latency. Not sure if Dynacabs or the shorter length IRs change the latency.

 

[paisleywookiee]

The USB recording function of the FM3 doesn’t report the latency of the blocks it seems. It reports the device latency, but I think Amp+Cab blocks add something in the 40ms range. If you search around on the Fractal forums, you’ll find threads on the topic.

I think it got fixed on the FM9 and Axe-FX III, but is still an issue on the FM3. I unfortunately couldn’t find a way to enter latency offset in GarageBand to automatically compensate for this problem.

What worked for me when recording and overdubbing with the FM3 as an audio interface on a computer or iOS device:
-Load up your typical FM3 patch with an amp and cab block. Pick a clean amp.
-Pick a test audio file with an identifiable transient like an IR.
-Send the audio file to Output 3+4 on the FM3.
-With the Input 1+2 Source in the I/O setting set to USB, record arm and record the reamped signal.
-Figure out the offset required to align your recorded signal with the test audio file.

On a full fledged DAW, I can figure out the exact number of samples to enter an offset into the DAW settings (somewhere in the 2100 range). On my iOS recording app (Cubasis), adjusting for 49ms gets it close enough for me. I think I typically have UltraRes IRs pulled up when checking this latency. Not sure if Dynacabs or the shorter length IRs change the latency.

Solid information, and mine would have been irrelevant anyway, had I not seen a squirrel and forgotten about this thread.

 

[bikescene]

Here’s the most informative thread about stuff not lining up with the FM3 USB recording:
https://forum.fractalaudio.com/threads/latency-compensation-measurement.177851/

Using the FM3, either through its analog outs or SPDIF, into another audio interface is a workaround for those who don’t want to mess around with offsets.

 

[paisleywookiee]

Here’s the most informative thread about stuff not lining up with the FM3 USB recording:
https://forum.fractalaudio.com/threads/latency-compensation-measurement.177851/

Using the FM3, either through its analog outs or SPDIF, into another audio interface is a workaround for those who don’t want to mess around with offsets.

I think the little Black Lion 2x2 has SPDIF, and works with iOS/iPadOS, which is a cool, albeit not cheap solution:

https://www.sweetwater.com/store/detail/Revolution2x2--black-lion-audio-revolution-2x2-2x2-usb-audio-interface

 

[metropolis_4]

@paisleywookiee @bikescene thank you both for the info!

Looks like I’m probably going to go interface shopping

 

[Sedaxel]

I´d swear that I read somewhere that Fractal units were quite bad regarding rountrip latency as audio interface.

 

[bikescene]

I´d swear that I read somewhere that Fractal units were quite bad regarding rountrip latency as audio interface.

I think the FM3’s reported latency is okay for a USB device of its age. I mainly use it with a buffer size of 128 and the I/O latency reported to the DAW is under 10ms at 4.1ms/4.5ms.

When the unreported samples are entered into the DAW as an automatic offset, the FM3 functions fine for overdubbing or reamping.

 

[Sedaxel]

I think the FM3’s reported latency is okay for a USB device of its age. I mainly use it with a buffer size of 128 and the I/O latency reported to the DAW is under 10ms at 4.1ms/4.5ms.

When the unreported samples are entered into the DAW as an automatic offset, the FM3 functions fine for overdubbing or reamping.

Yeah, but one thing is the DAW reported latency and a very different one (and much worse) is the real round trip latency.

You can easily measure it with a free app (RTL Utility), or alternatively recording a direct guitar at the same time that the processed output of that same guitar (with an splitter).

You'll find that real latency is, most probably, well above 10 ms (expect something around 15ms, if not more).

 

[Sedaxel]

In another thread, this was posted:

Post in thread 'FM3 as audio interface?'

Jan 6, 2024

Yeah this is what I don't know myself either. 12-13ms is acceptable if the buffer size is 512. It is not so acceptable if the buffer size is 64.

But nobody has actually said anything in detail when it comes to the FM3.

@GuitarJon what latency figures do you get for which buffer sizes? You (or anyone with an FM3) could measure them quite quickly using the Oblique RTL Utility here: https://oblique-audio.com/rtl-utility.php

I just measured the RTL of my FM3 using RTL Util. No blocks in chain other than input and output.

Out 2 L to in 1 L, FM3 USB buffer size at 48...

• aflynt

• 
• 

Not good RTL numbers for the FM3.

 

[Sascha Franck]

It reports the device latency, but I think Amp+Cab blocks add something in the 40ms range.

That can't be. It'd render the thing completely unplayable. Unless the latency only occurs on the USB stream, which would make no sense.

What I do however know is that there's been pretty noticable recording offsets with various FAS products. On a desktop computer you can usually adjust that offset in your DAW, on an iPad that's likely not possible.

Here’s the most informative thread about stuff not lining up with the FM3 USB recording:
https://forum.fractalaudio.com/threads/latency-compensation-measurement.177851/

Using the FM3, either through its analog outs or SPDIF, into another audio interface is a workaround for those who don’t want to mess around with offsets.

Ah, there we go.
How comes they haven't fixed that with a FW update yet?
Fwiw, typically, this isn't what is refered to as "latency compensation", it's a record offset (which obviously has to do with compensating some latencies as well - but in this case it's device latencies).

I think the FM3’s reported latency is okay for a USB device of its age. I mainly use it with a buffer size of 128 and the I/O latency reported to the DAW is under 10ms at 4.1ms/4.5ms.

From what I get from @metropolis_4 's post, I don't think he's using software monitoring, so audio interface latency isn't relevant - device latency however is, which needs to be reported to the DAW, so recordings line up properly.

 

[bikescene]

In another thread, this was posted:

Post in thread 'FM3 as audio interface?'

Jan 6, 2024

Yeah this is what I don't know myself either. 12-13ms is acceptable if the buffer size is 512. It is not so acceptable if the buffer size is 64.

But nobody has actually said anything in detail when it comes to the FM3.

@GuitarJon what latency figures do you get for which buffer sizes? You (or anyone with an FM3) could measure them quite quickly using the Oblique RTL Utility here: https://oblique-audio.com/rtl-utility.php

I just measured the RTL of my FM3 using RTL Util. No blocks in chain other than input and output.

Out 2 L to in 1 L, FM3 USB buffer size at 48...

• aflynt

• 
• 

Not good RTL numbers for the FM3.

Yeah, I agree that it’s not great when overall RTL is considered. I do think it’s something that Fractal should try to improve on if possible.

That can't be. It'd render the thing completely unplayable. Unless the latency only occurs on the USB stream, which would make no sense.

What I do however know is that there's been pretty noticable recording offsets with various FAS products. On a desktop computer you can usually adjust that offset in your DAW, on an iPad that's likely not possible.



Ah, there we go.
How comes they haven't fixed that with a FW update yet?
Fwiw, typically, this isn't what is refered to as "latency compensation", it's a record offset (which obviously has to do with compensating some latencies as well - but in this case it's device latencies).



From what I get from @metropolis_4 's post, I don't think he's using software monitoring, so audio interface latency isn't relevant - device latency however is, which needs to be reported to the DAW, so recordings line up properly.

I’m sure why they haven’t fixed it yet. I occasionally check and it appears to be a low priority item. You are right about it being overall device latency. Maybe passable if he is recording 1 track over a backing. But a little more of a problem if he’s overdubbing another track to lock in with the first one.

 

[Sedaxel]

Last night I was watching Ola´s review of the Valeton GP200, and he talked about 5ms latency being unusable for him live. The unit has those 5ms RTL without any block engaged, but it´s hard for me to believe that it´s too much, considering it´s more or less what units have with a complete signal chain. Maybe with effects, amp and cab, the Valeton number raises to 7-8 ms, and then it becomes noticeable.

Personally, I don´t really notice 8 ms. But I certainly feel those 13-15 ms of the FM3 (as audio interface, I mean). It would be unuseable for me for playing guitar in that setup.

 

[LP59]

Last night I was watching Ola´s review of the Valeton GP200, and he talked about 5ms latency being unusable for him live. The unit has those 5ms RTL without any block engaged, but it´s hard for me to believe that it´s too much, considering it´s more or less what units have with a complete signal chain. Maybe with effects, amp and cab, the Valeton number raises to 7-8 ms, and then it becomes noticeable.

Personally, I don´t really notice 8 ms. But I certainly feel those 13-15 ms of the FM3 (as audio interface, I mean). It would be unuseable for me for playing guitar in that setup.

I don't think Ola was talking about RTL. He was talking about the processing latency. For the FM3, that's about 4 msec. I believe the Valeton GP200 is about 3 msec.

The RTL usually doesn't matter because you monitor direct when recording. The only problem is if the RTL isn't reported to the DAW correctly. Your recording will be off in that case. Like Sascha says, most DAWs have a parameter you can set to compensate for that.

I watched Ola's video and he's wrong about what he's measuring. He thinks he's measuring RTL, but he isn't. He's measuring the compensation error. In some cases that happens to be close to the processing latency, but not always. That will never be the RTL though. It looks like he's been doing those measurements for at least a couple of months now and showing them in his videos like he's measuring something significant. Somebody should probably explain this to him.