The thing is, you don't have to do anything complicated. You plug in your interface, install the driver, choose that driver in your software and you're ready to go. Compensation will be done automatically. Your recordings will be in sync, you can increase the buffer size when mixing if your CPU peaks in the red...

 

Last edited: Oct 24, 2024

That's no resistance, that's physics. You misunderstand the concept of latency and it's compensation. Any digital system needs some time to calculate audio in blocks. You can reduce block size, but thus raise computing power. And there are some factors like AD/DA converter delay and buffers you can't get to zero.

It's just like you are asking why it's not possible to make a starter gun in a distance of 100 meters sound the same time as it flashes.

 

Last edited: Oct 24, 2024

I didn't realize that the measurement of 729 at 96k could be less latency in terms of time (ms) than 429 at 48k.

 

Last edited: Oct 24, 2024

You have to understand that you can't completely compensate something that is already there. Like if you insist to have a piano thats playing a full level note as soon as your fingertip touches the surface of a key, eliminating the delay of hammer action, striking the strings and finally the time it takes to get to your ears. If you don't understand it this way, I don't know how to explain it anymore. Latency is part of our hearing, our surrounding, instruments, speed of sound at 343 m/s is awfully slow, so 10 ms of latency should be perfectly fine.

On the other hand, imagine a drummer who hits the snare and is used to have it 0.5 meters away from his ears, that's about 1.5 ms latency. Now let him play a drum VSTi over MIDI with 10 ms MIDI delay and an audio interface with another 10 ms latency, then the snare sounds like it is about 6.9 meters away.

You can try to get a better controller to reduce MIDI latency, and to reduce audio interface latency, but there is no way to completely eliminate it. And how low you can get it and with what latency the drummer can get comfortable to play is not a matter of zero or not, it's always a compromise.

 

It's encoded in the term itself.. "sample rate"

44100 sample rate = 44100 samples per second
48000 sample rate = 48000 samples per second
96000 sample rate = 96000 samples per second

In other words, 1 sample at 48000 takes twice as long as 1 sample at 96000

Or to plug in your numbers from earlier:

96000 -> 729 samples
48000 -> 429 samples

(1 / 96000) * 729 = 0.00759375 = 7.6ms latency
(1 / 48000) * 429 = 0.0089375 = 9ms latency

So your interface incurs ~1.4ms less latency at 96kHz compared to 48kHz.

E: Oops, Myfanwy was first - I only just now started reading earlier posts, sorry.

 

Wow! I did not realize this. I probably used to know it, but forgot. LOL.

Are you saying that recording at 96k with a 729 reported latency in samples would actually result in less millisecond latency than 48k with a 429 reported latency in samples? This information is very useful. Thank you.

 

Having low audio latency is a must. If you are connecting directly into the streaming computer make sure the buffer settings on your interface are as low as possible (often 32 or 64 samples). Make sure that the sample rate is set to either 44.1 or 48khz as live streams won’t be at higher sample rates.

Audient - iD14 MKII - Manual
Set Sample Rate Sets the operating sample rate of iD14. 44.1, 48, 88.2 and 96kHZ are the four options available.

Set ASIO Buffer Size
Sets the buffer size of the iD14 between 16 and 4096 samples. Higher sizes will take processing load off your computer but will cause increased latency

 

Admittedly, this is more complex than I imagined; however, knowing that latency can be reduced . . . I will reduce and nobody can convince me not to. My loopback tests and RTL Utility results and my DAW off-set suggest that I can get to ZERO. Even if that's not sustainable throughout a recording process as it's fluid, why not get close to ZERO as opposed to saying, "It's good enough."????

 

As we age, our ability to hear higher frequencies diminishes, but the human hearing range typically spans from 20Hz to 20kHz. To capture the complete range of sounds audible to humans, sample rates of 44.1 and 48kHz are more than sufficient.

As such, the vast majority of digital music available by typical distribution methods (streaming on Spotify/Apple Music, CDs) is at a 44.1kHz sample rate, and audio for film tends to be at 48kHz3.

What's the point of higher Sample Rate options?

Since sample rates of 44.1/48kHz allow us to capture frequencies spanning the full range of human hearing, you wonder what the purpose of higher sample rate options is.

There is debate in the audio community about the value (or lack of) of using higher sample rates for situations that don't fall into the above categories (I.e., for general recording purposes). We won't get into that here…

 

Still some way to go...

Nobody wants to convince you to do anything, just to understand how it works, and there are two fundamental differences.

You don't have to mess with DAW offset settings at all if the driver developer of the audio interface did everything right. Latency for recording is already compensated so that you have virtually zero latency, no matter what buffer size you are using.

What you can't compensate is the real time round trip latency, that's the time it takes to process audio from analog in through the DAW to analog out in real time. You only need that for live monitoring of any signal through an amp sim for example. Most singers prefer a full analog monitoring setup as I stated before, more due to comb filtering in the hearing than latency. Most keyboard players are already dealing with latency due to MIDI or mechanical action. Compensating this would literally mean time travelling.

Edit:

Simplified, if you play back audio to headphones and sing along into a microphone without giving the microphone signal to headphones at all and just move the headphones away from one ear, and record this, it will be perfectly in time with the playback, no matter what buffer size, sample rate or latency your audio interface has.

Same if you give the microphone signal purely analog over a small mixer (maybe integrated in your audio interface) together with the DAW output to your headphones.

Now if you decide to use software monitoring and process your voice with some EQ or compression in your DAW and give this processed signal to your headphones, you introduce round trip latency. And this is the only case you have think about latency.

For sure it is the same case using any real time plugins while playing like guitar amp sims or VST instruments, as they rely on real time latency.

Hope this helps.

 

Last edited: Oct 24, 2024

So tommyzai, here are two pictures for you (it's Studio One 7): 32 Bit Float | 48.0 kHz | 64 Samples

 

Last edited: Oct 26, 2024 at 11:45 AM

With that number in mind, now consider the fact that the average human reaction time is 250 milliseconds.

Unless your processing a signal for live performace Loop tests are meaningless.

What your recording you monitor in parallel to the input, not from the computers output. If not how would anything ever have got recorded prior to 2005 when anything less then 5ms was see as good.

If your recording a guitar on a marshall amp, you listen to the amp the guitars plugged into when recording. Not the live monitor play back on the computer connected to mic faceing the amp, which is the signal path your mesuring the time of when you measure loopfeedback.

Legotron was right when he said

Any delays will get fixed when you quantise to your click track anyway.

When your singing into a mic, to process with plugins, to play to a live audence with a live band. Thats when loopfeedback matters.
When recording, it's meaning less other then to test how powerful a system is.

 

After all of my loopback latency experiments, I fear I have things set up incorrectly. I have been turning up the level using my headphone knob and cannot turn up my channel knob. Something is wrong. I get feedback when I do . . . even with monitoring off on DAW and Interface. What am I doing wrong? ;-0

 

tommyzai please go to:
What is Audio Loop-back and how to use it?
https://audient.com/loopback-made-simple/

 

LOL!! Thanks. I had no idea my interface had this feature. It works well with an internal signal. But I'm thinking maybe this doesn't account for going out the headphone jack. The latency reported doing a Loop-back test this way resulted in 1/2 the latency measurement of the other methods. I realize we are talking about tiny ms, but not sure what amount to off-set.

 

Last edited: Oct 25, 2024 at 6:13 PM

Does this RTL Ultility Report indicate that my interface is over-reporting by 54 samples? That's how I'm reading it, but . . .

Reported: 429
RTL: 375/7.812ms


Would latency measurements change with each DAW, or would it remain constant?

 

One final question . . .

If . . .

Interface Reports 400 samples latency
RTL and all other tests indicate 300 samples latency

Would you . . .

Check Use Audio Driver Reported Latency and off-set by -100

or

NOT check Use Audio Driver Reported Latency and off-set by +300

I have been checking and off-setting a negative measurement, but then I thought . . .

Is that creating one extra process for DAW? Does DAW have to deal with the reported and then compensate as opposed to just compensating?

 

Sometimes, settings on your PC can stop you getting the best performance, so we’ve written a handy guide to help you optimise your PC into an audio processing beast. You can find that guide here.

How to Fix Latency in REAPER (Easy to Follow Guide!)

 

Last edited: Oct 26, 2024 at 2:41 AM

The latencies depend on many things such as the performance of your computer, your operating system, the number of mid and audio tracks and the plugins used, so values cannot be transferred to other PCs.

 

Last edited: Oct 26, 2024 at 2:53 PM

Hey "Mr. copycat" aka BEAT16, posting Windows tutorials won't help a macOS user like the OP very much.

And tommyzai, please go through the whole thread again and read carefully, every information is already there.