44.1khz master only SPDIF, latency and the Kemper

correct me if I'm wrong (and this might be the case here) but I always thought the higher the sample rate the more cpu usage. So if that's right a higher sample rate wont save you from dropouts you otherwise get from a lower buffersize.
And these days most machines can handle a lower buffersize .... if you're not using 20 instances of your plug ins on every channel while tracking

Quote from nightlight

Was reading a Sound on Sound article and came across this interesting line: "For instance, in the case of a typical ASIO buffer size of 256 samples in a 44.1kHz project, latency is 256/44100, or 5.8ms, normally rounded to 6ms. Similarly, a 256-sample buffer in a 96kHz project would provide 256/96000, or 2.6ms latency."

This interested me greatly, because my PC is not really souped up and I've been forced to use higher buffer settings in order to avoid dropouts.

I know we've discussed the merits and demerits of the Kemper's SPDIF "master only" policy with 44.1 khz and I understand the logic behind the decision. That said, given that recording in higher sample rate does offer the benefit of reduced latency, perhaps this is one other facet that I feel should have been taken into consideration while deciding the sampling rate.

Thoughts, anybody?

Display More

We could have a option for this in the system menu, faster is bether. if you dont want all the fx on and just use it as an amp I think this is possible.

256 samples is the default value for many interfaces, but with a well configured DAW you can easily go way below this default value. My DAW for example is running at 48kHz and the audio interface is set to 32 samples buffer, no issues.

Regarding your PC ... I'm not sure you would have success using 96kHz with 256 samples buffer, because that would mean you can use 44.1kHz with 128 samples buffer as well. Or the other way round ... if you NEED to use 256 samples with 44.1kHz, then you would need to set the buffer to at least 512 samples when switching to 96kHz.

Also keep in mind that DAWs should be able to automatically (or through custom setting) compensate for latency. So I don't think, latency is a reason to go for 96kHz. The only reason for using 96kHz I can think of is that you can slow down audio footage without compromising the audible frequency spectrum (Nyquist). A rare use case for most, but in audio effects world this CAN be helpful sometimes. Record an explosion and slow down the "BANG" to make it more "BOOM" without loosing the higher parts of the recorded frequency spectrum. Very theoretical example, I know. Difficult to record usable frequencies beyond 20kHz with common microphones and pre-amps. But I think it serves well for explanation.
96kHz is a rarely required sample frequency ... 48kHz is a different story for different reasons though.

Quote from nightlight

"For instance, in the case of a typical ASIO buffer size of 256 samples in a 44.1kHz project, latency is 256/44100, or 5.8ms, normally rounded to 6ms. Similarly, a 256-sample buffer in a 96kHz project would provide 256/96000, or 2.6ms latency."

This conclusion can be misleading. A 128-sample buffer in a 44.1kHz project also has 2.6ms latency. Typically, ASIO buffers are 64 or 128 today. This increases the "stress" for the system since it increases the overhead that comes with the buffer handling (not the audio rendering).

Nevertheless, even if the buffer size is 128-samples, usually bus and driver handling can add multiple buffers, since jittering or even out-of-order data need to be intercepted with multiple buffers.

Since the overhead of handling small buffer sizes is quite a number, VST/AU hosts usually render one track only, the "selected" track or the ones that are recording. All other tracks are rendered with larger buffer sizes (and "not so" realtime), but since the musician does not interact with them, he/she doesn't notice it.

Well, the computer devoted to making real-time music should be optimized for the purpose: killing most of the services, no antivirus, no firewall, no internet, no background tasks, the minimum amount of drivers.
I use a dual-boot option for this.
Also, mandatory IMO are a motherboard optimization (BIOS), firewire or internal ASIO soundcard with 2 ms latency declared, excellent drivers and direct monitoring option.

When this is not possible, Windows for example (I use a PC) allows to create different users with different environments (drivers, services etc): it's not the same, but it helps.

Quote from nightlight

... I migrated to Samplitude because I prefer the "sound" (entirely subjective, but I'd say it has a more open and neutral character) ...

Don't get me wrong, I'm just joking but this is the second weirdest statement ever, right after "the earth is a flat disc".
On a more serious side ... if you prefer the filters for some reason, ok. But the software flat in flat out can't make a difference in sound.

yeah, you should be able to simply copy any vst plug-ins from one DAW software to another.

Quote from nightlight

Was reading a Sound on Sound article and came across this interesting line: "For instance, in the case of a typical ASIO buffer size of 256 samples in a 44.1kHz project, latency is 256/44100, or 5.8ms, normally rounded to 6ms. Similarly, a 256-sample buffer in a 96kHz project would provide 256/96000, or 2.6ms latency."

You need to double that amount. That buffer number applies to in and output buffers.

I'm curious though...are you monitoring the Kemper through your DAW when recording? Is there no mixer for your audio interface for direct routing? That way you can use whatever buffer you'd like with no latency.

Quote from Trazan

Is there no mixer for your audio interface for direct routing? That way you can use whatever buffer you'd like with no latency.

Sure there is, MixControl is pretty powerful once you get the idea of the custom mixes and the routing. Latency shouldn't be an issue at all.

EDIT: Oops, I mixed up two threads, sorry. The OP seems to have a Fireface instead of the Focusrite Scarlett I assumed. But TotalMix should have similar functionality. Direct Monitoring for sure.

A double sample rate does not result in a lower latency.
It results in putting double calculation power into the same musical output.
You might end up increasing the latency to get a back a bit from the "sactified" calculation power.

44.1 kHz provides the lowest load for your computer, thus the highest probability that a low latency setting will work for your project.

Quote from nightlight

I'm no audio expert, but that's what the Sound on Sound article suggested, so I thought I'd post. Perhaps it's not exactly a 1:2 difference, but I'm sure there must be some correlation.

Yes, you get lower latency at a certain sample buffer since each sample is shorter. I think CK means that if processing power is what dictates usable buffer size you won't gain anything since higher sampling rates requires more processing power...and you'd have to raise the buffer.

You just divide the sample buffer with sampling rate:
256/44100 = 5,8 ms
256/96000 = 2,7 ms

As to having to use a 1024 buffer on your system....something must be wrong. It should definitely be capable of doing better than that, unless you're doing humongous projects..

Quote from nightlight

You'd be surprised at the number of people that say this about Samplitude. All boils down to algorithms, just like the Kemper vs Axe FX II in some ways. Cubase sounds darker, Samplitude a little more open, imho.

Statements like this have been tested and disproved on many different audio forums. It's likely the default pan law is the culprit, unless you are using different plugins which isn't a fair comparison. In any case, it's a modern day 'flat earth' analogy...laughable. My audio teacher swore Pro Tools sounded worse than DP, likely a different Pan Law was influencing him as well......

The Axe and Kemper are WAAAAY different beasts. The closest A/B test you could do on them is to profile the Axe with the Kemper. But I wouldn't be surprised if there were subtle differences - profiling a digital amp model is VERY different than making a digital copy of something. And really that's all DAW's are doing - they are digitally mixing tracks. The algorithms should be equally (100% transparent). There is no impact on frequency response or other aspects of tone, until you add effect sends.

If you are inclined, try a double-blind test. Put the exact same audio tracks in both DAWs and export the audio mixdowns .wav files. Label them track 1 and track 2, so if someone else saw them, they wouldn't know which file came from which DAW. Then have a friend/roomate/etc play each file in a random order (ex. 1,1,2,1,2,2,2,1,2) and ask you which DAW you think it came from. Make sure you do this enough times to make the correlation statistically relevant. I think 10-20 attemps is enough. Then look at the recorded results and score how many times you picked the right DAW. I'd be surprised if you score better than 60%.

I just remembered Samplitude has an additional audio engine mode for low-latency monitoring, I found this info with a quick google:

Quote

as you already pointed out, hybrid engine is especially for monitoring through the software, when you need low latency.
in this scenario, you can lower the cpu consumption, by switching channels, that are only playing back audio to economy engine, where they are played with longer buffers.
for mixing, the economy engine is stable and good (slider to the left in settings).
one thing that is maybe interesting is, that in hybrid where the audioengine is always running, you hear the tails of reverbs and other effects that prolongue sound, when you hit stop. can give some good information, what is going on there.

BTW I use a Fireface 800 as well and it's perfect for no latency routing, have a look at that for a solution. I can usually get down to 48-64 if I need to in Reaper, although 128 is my default and I just use TotalMix to monitor inputs with zero latency.