From what I've read on the various forums around the traps, JSB's experience is pretty much what everyone else sees too. Increasing a DAW's SR results in lower latency at any given buffer setting, and the conventional reasoning is simple:
Buffer sizes are measured in samples. If the rate at which they stream through software is doubled, you'd expect roughly a halving of the time it takes for any given number of them to pass through.
I think this is a classic case of two people talking about different things. JSB is talking about latency when using plugins and monitoring through his DAW (ie having a channel strip active while recording vocals, or playing midi instruments live), whereas Mr Kemper, I believe, is talking about latency in the KPA. JSB is correct, as is your statement, Nicky. CKemper is undoubtedly right if talking about the KPA, too.
Indeed, I suspected that, which is why I chose not to mention CK, Sammy.
I was responding to JSB and Harry's posts.
Hope you're well, brother. 
...The overall latency in the DAW was what I was concerned with. If I use 96KHz, I get 3.96ms overall latency, when monitoring in real-time. If I use 44.1KHz, I get 13.1ms overall latency.
In the pictures you showed, you might want to click on "Hardware setup" or input setup. I bet when you switch from 96k to 44. the buffer size increases and that's why you get more latency. If you pick a buffer size of 128 or 96 or 64 you should get the low latency at 44.1.
In the pictures you showed, you might want to click on "Hardware setup" or input setup. I bet when you switch from 96k to 44. the buffer size increases and that's why you get more latency. If you pick a buffer size of 128 or 96 or 64 you should get the low latency at 44.1.
Read the paragraph titled "Value Judgments"
Click!
It's to fix some clicking issues when connecting via spdif with sound cards of a certain hardware configuration, apparently. It only affected a very small number of users, but Kemper's policy is a bit like catering for lactose-intolerant kids at a birthday party by making two cream cakes; one with real cream and one with soya
Go Kemper!
I don't have all of my set-up details, but I'm still getting some clicking when using SPDIF out into my Focusrite 6i6. I *think* I have it set to 44.1 in the configuration but I'll double check.
In the pictures you showed, you might want to click on "Hardware setup" or input setup. I bet when you switch from 96k to 44. the buffer size increases and that's why you get more latency. If you pick a buffer size of 128 or 96 or 64 you should get the low latency at 44.1.
I double checked to make sure the sample sizes were set to 128, for both 44.1/96KHz; a 10ms second Overall-latency difference, definitely sounds extreme, but those are the correct numbers being shown.
Also, thanks to Monkey_Man and sambrox, for more accurately presenting what I was trying to say.
.
I bet when you switch from 96k to 44. the buffer size increases...
I don't know the DAW, Dean, but buffer size in this context is always a chosen number of samples, in this case 128. As I said earlier, the rate at which those samples pass through the software will determine the delay experienced.
IOW, changing the sample rate has nothing to do with the chosen number for the buffer, only how-quickly the data is able to pass through the system. This increased speed of data flow obviously taxes the CPU/s more 'cause they have to process more data in the same timeframe.
Higher SR = shorter pass-through time = lower latency & higher CPU load.
.
I agree Nicky but the way some drivers present this to DAW can be different from different hardware audio interface makers. You're science is correct, but the hardware limitations usually entails that certain low buffer settings are many times not available at higher sample rates so in practice, you generally will have a better chance of getting a functional lower latency performance at the lower sampling rate and not the other way around. If the Focusrite is operational at these low latency numbers at 96khz, that would be amazing.
Not so long ago, it was really expensive to build, or buy a DAW that was capable of solid multi-track recording at 96/192KHz. These days, any i7 PC or Mac should have no problems, especially if using SSD's.
The 2nd Gen Focusrite drivers and hardware, were a huge step up from my 1st Gen 6i6. I have the same PC and DAW software, I changed only the interface, the 1st gen 6i6 would have an overall-latency of 7.65ms at 96KHz, whereas changing the interface and drivers, the new 18i8 is 3.69ms at 96KHz. I always use 128 samples, as that seems to be the sweet spot with my PC hardware.
The 2nd Gen Focusrite seems to be a major improvement. If the KPA could do 96KHz over SPDIF, I'd have nothing left to ask for.
You could still use a sample rate converter. Wouldn't affect the sound and you could always use the same sampling frequency if that is important for you.
I think this is a classic case of two people talking about different things. JSB is talking about latency when using plugins and monitoring through his DAW (ie having a channel strip active while recording vocals, or playing midi instruments live), whereas Mr Kemper, I believe, is talking about latency in the KPA. JSB is correct, as is your statement, Nicky. CKemper is undoubtedly right if talking about the KPA, too.
In my post I was referring to the latency of DAWs and audio interfaces, not the Profiler.
The lowest possible latency will typically not be achieved by choosing a higher sample rate, but by trying to lower the blocksize to a minimum at a given sample rate.
There is also some words to be said about the usage of 96 kHz.
We do not recomment to use it because it has no audible effect on proper equipment and software.
It was audio scientists who invented digital audio and the equipment.
It was the equipment industry that introduced 96 kHz by the request of users, although not driven by the science.
96 kHz is sometimes recommended by audio journalists and music producers.
In contrast, no serious audio scientist recomments it.
It is also interesting to observe that no equipment manufacturer (DAW, audio interface) that provides 96 kHz, recomments to use this valuable feature for superiour audio results.
Why is that?
I work for TV as a Sound Engineer and over all those Years
everything works with 48 kHz and nobody is trying to change that
no need for 96 kHz
As the saying goes, "Bigger is always better".......... Not
On the few projects I have mixed at 96k, it is hard to know if you are just trying to convince yourself it sounds superior. I do think that plugins perform better, but in the end there is more to be gained by concentrating a little more time on the mix.
I too work as a sound engineer in TV. The thing is, even though we all use digital audio everyday, whether it's lossy formats on your iPhone or listening to digital radio, no one seems to want to understand the science behind it, hence the common misconception that a higher sampling rate equals smoother curves on a sine wave or higher resolution as with HD TV. At a basic level, in order to capture a particular frequency, the sampling rate has to be twice that frequency, in order to capture the peak and the valley. Therefore, in layman's terms, a rate of 44.1 kHz can capture frequencies up to 22 kHz (there is always a little leeway to accommodate the low pass filter's resonance when filtering out frequencies above the ceiling, which is used to stop the unwanted higher frequencies being "folded back" erroneously in to the digital audio information as lower frequency distortion, known as aliasing). As the human ear can theoretically hear a maximum range of 20 Hz to 20 kHz (in most cases quite a bit less), a sample rate of 44.1 kHz should suffice for any application, though there are arguments for higher rates if the filters and any dithering aren't up to scratch (dithering is one way to deal with any extra ultrasonic frequency information that might otherwise cause aliasing, by placing it back outside the audible spectrum, or at least high up enough that it won't be so noticeable). By the way, it's no coincidence that cinema runs at 24 frames per second and the broadcasting standard sample rate is 48 kHz...
.
I thought dithering was only intended for bit rate conversion and has no use for sample rate conversion. Am I wrong?
Haha no you're right! I started writing that post driving home from a trip to the countryside. My phone died half way through, and when I put it in a charger and restarted it when I got home, the post was gone. I hit reply and the text reappeared, though by this time I'm afraid the afternoon beers had made me forget exactly what I was writing about, hahaha! That's pretty embarrassing considering I was getting all holier-than-thou 
