You guys done yet?

 

I think and I hope so, Xorome! Thanks for asking!

 

You can't eliminate latency in digital. Knowing the exact total RTL of your setup can be useful for many things but you can only reduce latency by decreasing the buffer size and/or increasing the samplerate of your interface.

 

That's it, you have to live with latency. But it's helpful to know the exact specs of your setup and what you really need to work without distraction.

And that's where the problem starts, if you have not noticed any lag that distracts you while playing, why bother?

Edit: A simple example:

A delay of 7 ms equals 2.4 meters distance (assuming speed of sound is 343 m/s). If you stand 2.4 meters away from your guitar amp, is it already distracting you while playing or feels laggy? Can't believe that.

 

Last edited: Oct 24, 2024

When I raised from 48k to 96k the latency jumped way up.

 

Should be the other way round. What do you mean by "jumped way up"? The reported latency in Reaper or any tool like RTL Utility? Or the perceived latency while playing?

It would also be good to know what you are recording. Guitar, bass, keyboard, drums, other instruments, vocals?

 

I almost think you want to solve a problem that's simply not there, just because you read about it and want to get it to zero, which isn't possible in any digital environment.

Latency isn't about audio quality or incorrect placement of recordings, it's only about live monitoring while playing or singing and distracting your performance from correct timing and feeling.

 

I tested with RTL and a manual track test a la Kenny Gioia . . . comparing samples on two tracks . . . one loopbacked. These tests have been line in.

Latency is real and present. It can be measured and compensated. I would consider myself a lazy hack if I didn't take the time to improve my recording. ;-) I want as close to zero as I can get. I certainly don't want any lag when monitoring and playing/singing. Note: I have not noticed anything. I am merely doing this as due diligence to set things up as best I can. Why not? What's the rationale for living with latency when there is no need. It's like the cabinet maker who doesn't bother fine sanding the edges.

 

I think you don't improve your recording, you are just overcomplicating it. You don't have to care about latency compensation or RTL if you don't actually play an instrument or sing.

So again, do you even play an actual instrument or VSTi or sing? Or are you just testing in hope to get everything setup perfectly before starting to make the next world hit?

 

Because you simply can't.

If it's your goal to get real zero latency, feel free to buy an analog mixer, analog tape recorder, analog speakers without DSP and use only analog hardware like EQs or compressors. Use headphones or glue your speakers directly to your ears, because a distance to the speakers of 1 meter gives you 3 ms of latency.

And also only analog instruments, no MIDI, no digital Synthesizers, not even ones with digital controlled oscillators or VCAs. Good luck.

Edit: Like in my example above, every guitarist standing 2.4 meters away from his cab has to deal with 7 ms of latency. If he would insist to compensate that, he could put a microphone directly to the speaker and wear headphones or lie on the floor with his head touching the cab.

Just accept that latency is perfectly normal and can never be absolute zero.

The only situation where it really matters is vocal monitoring, because you are hearing your voice inside your head directly through your bones and also over headphones. Any delay including the distance to the microphone will introduce comb filtering which may be distracting for singers.

 

Last edited: Oct 24, 2024

Maybe because the minimum buffer size of your interface at 96k is 128 samples. If so, 48k @64 samples probably gives you the best low latency possible with this interface.

 

Audient advertise their V4 drivers being capable to run with 16 samples buffer size at 96 kHz, but it's indeed important to check all settings before comparing anything.

 

You seem to imagine that offsetting/compensating to perfectly align your digital system with the outside world will eliminate latency...

 

I truly appreciate half of the comments. Thank you. ;-) Do I actually sing? Play an instrument? Record? Well, I am in the process of setting up this DAW to do just that. Shouldn't I assemble my studio first? The goal is to determine the reported latency and compensate for it. I really don't understand why the resistance to latency compensation. To me it's natural to set the studio the very best that I can. Living with latency that can be corrected is like not bothering to clean tape heads, using faulty mic cables, running a loud A/C unit in the vocal booth, etc. I'm simply trying to get it all set up the best I can in preparation for recording.

 

And that would be time travelling. Any musician is dealing with latency every single day, whether it's a guitar player in front of his cab, a piano player dealing with the latency of hammer action, or the "holy grail", a church organ player, who even has to compensate notes differently due to the slow build up of low notes.

 

What the? I'm an educator, lyricist, musician, and producer. Are you being nasty and condescending or are you asking if I will be using live vocals and instruments as opposed to MIDI?

 

Would you go with 48 or 96k in this case (as per RTL Utility). I should probably see how my interface does with reporting and correcting on its own. At 48k it over-reported and compensated by 62 samples.

Audient iD14 96000/64 729 (reported latency)
Audient iD14 48000/64 429 (reported latency)

 

Please don't misunderstand me. But as an experienced musician you should know about speed of sound, attack behaviour of instruments and it's compensation any musician does while playing, like described above.

Your brain can compensate and play a low key on the church organ earlier, but how should this be compensated in a digital environment after the note has been played?

 

I still do not understand the resistance. If latency can be compensated . . . getting closer to zero, why not? Makes no sense to me that one would say, "Well, I can't get it perfect (zero), so why bother?"

 

That's simple, 729 samples at 96 kHz equals 7.59 ms, 429 samples at 48 kHz equals 8.94 ms, that's about 0.5 meters in distance.

Not much of a difference in real world, but running your session at 48 kHz needs much less CPU power than running it at 96 kHz.