Audio interface suggestions needed! (Update: RME interface has arrived!)

octatonic

goldtop said:

I think you'll be happy. I came very close to buying a Babyface, but ended up with a UFX (only because I hate plugging and unplugging stuff and cable clutter).

When I saw that RME reprogrammed the FPGA in the original Babyface to add a Class Compliant mode for use with the iPad, I knew they were the sort of company that looked after their existing customers.

RME support is the best in the business.
They've are still updating software for products no longer being made.
That almost never happens.

Branshen

goldtop said:

I think you'll be happy. I came very close to buying a Babyface, but ended up with a UFX (only because I hate plugging and unplugging stuff and cable clutter).

When I saw that RME reprogrammed the FPGA in the original Babyface to add a Class Compliant mode for use with the iPad, I knew they were the sort of company that looked after their existing customers.

That's one of the reasons I got it. Even though it's USB 2.0, I'm not really worried about them not supporting it in the future because of their reputation. 

Branshen

Took nearly a month, but the RME has arrived (import-duty free)! I will set it up and give some feedback on how it compares to my Scarlett.

oaf

:-) hope you like it, think you will!

Branshen

Had a good go tonight. First thing I noticed was that it is alot better than my old 2i4 which struggled to play large projects without crackles and constant dropouts. (I had to devise a workaround and export a master track everytime I loaded a session for mixing to stop the crackles) 

Latency is sub 5ms when the buffer samples is at 64, which is awesome to play guitar through VST amps and effects. but this is only possible for empty or near empty projects. larger projects will start to crackle at this buffer sample size. So it will not do miracles. 

I guess the bottleneck is my laptop, and not my interface now. I notice that the crackle happens at low buffer samples and the CPU usage hits around 50% in ableton. 

octatonic

Branshen said:

Had a good go tonight. First thing I noticed was that it is alot better than my old 2i4 which struggled to play large projects without crackles and constant dropouts. (I had to devise a workaround and export a master track everytime I loaded a session for mixing to stop the crackles) 

Latency is sub 5ms when the buffer samples is at 64, which is awesome to play guitar through VST amps and effects. but this is only possible for empty or near empty projects. larger projects will start to crackle at this buffer sample size. So it will not do miracles. 

I guess the bottleneck is my laptop, and not my interface now. I notice that the crackle happens at low buffer samples and the CPU usage hits around 50% in ableton. 

Realistically 128 samples is normal for most people.
64 is only really for this people with massively powerful computers and only at the start of the session.

Freebird

octatonic said:

Realistically 128 samples is normal for most people.

64 is only really for this people with massively powerful computers and only at the start of the session.

I always track @ 64 samples, even if I have to freeze and put the VST's offline, which is a simple one-click affair in Reaper. I sometimes DI my electric guitars and bass, so that I can hear the interaction between each instrument more clearly, and then I reamp them later. This has the side effect of not stressing the CPU (same if you use a Kemper/Helix/Axe FX). Finally, after every track has been converted into a stem, I play around with the Nudge tool to see if I can improve the overall feel and groove.

Branshen

octatonic said:

Branshen said:

Had a good go tonight. First thing I noticed was that it is alot better than my old 2i4 which struggled to play large projects without crackles and constant dropouts. (I had to devise a workaround and export a master track everytime I loaded a session for mixing to stop the crackles) 

Latency is sub 5ms when the buffer samples is at 64, which is awesome to play guitar through VST amps and effects. but this is only possible for empty or near empty projects. larger projects will start to crackle at this buffer sample size. So it will not do miracles. 

I guess the bottleneck is my laptop, and not my interface now. I notice that the crackle happens at low buffer samples and the CPU usage hits around 50% in ableton. 

Realistically 128 samples is normal for most people.
64 is only really for this people with massively powerful computers and only at the start of the session.

Gotcha. Still a great unit regardless. There were so many weird quirks with my scarlett that I'm happy to be rid off (i.e. needing to hotplug it everytime I restarted the computer). Really bad crackling+dropouts for large projects. And crucially, I had to avoid using a good vocal take in a recent recording session because the 2i4 somehow managed to record a crackle. No idea where it came from but that's one good take ruined. 

Onboard reverb is a game changer!

Will have to figure out a problem where chrome doesn't want to play nice with the babyface, seems to be chrome bug that's materialised itself, as many others have encountered it too..
https://support.google.com/chrome/thread/2134843?hl=en

stratman3142

Branshen said:

Latency is sub 5ms when the buffer samples is at 64, which is awesome to play guitar through VST amps and effects. but this is only possible for empty or near empty projects. larger projects will start to crackle at this buffer sample size. So it will not do miracles. 

I guess the bottleneck is my laptop, and not my interface now. I notice that the crackle happens at low buffer samples and the CPU usage hits around 50% in ableton. 

So far my current PC has been able to cope with anything I throw at it, even at 64spls. However, on my previous, less powerful, PC I had to make various adjustments to allow lower spls settings for tracking without the dreaded pops and crackles. You can Google how to optimise your PC for audio. Apologies if you already know this.

IIRC, the one that really helped on my old PC was to "Adjust for best performance", which could be accessed as follows:

Control Panel > System and Security > System > Advanced system settings > Performance > Settings

Then select "Adjust for best performance"

I think I ended up with a pretty basic grey looking display in Reaper, but at least it allowed me to record with lower latency. Then afterwards I could set it back to the setting that says "Let Windows choose what's best for my computer"

Branshen

stratman3142 said:

Branshen said:

Latency is sub 5ms when the buffer samples is at 64, which is awesome to play guitar through VST amps and effects. but this is only possible for empty or near empty projects. larger projects will start to crackle at this buffer sample size. So it will not do miracles. 

I guess the bottleneck is my laptop, and not my interface now. I notice that the crackle happens at low buffer samples and the CPU usage hits around 50% in ableton. 

So far my current PC has been able to cope with anything I throw at it, even at 64spls. However, on my previous, less powerful, PC I had to make various adjustments to allow lower spls settings for tracking without the dreaded pops and crackles. You can Google how to optimise your PC for audio. Apologies if you already know this.

IIRC, the one that really helped on my old PC was to "Adjust for best performance", which could be accessed as follows:

Control Panel > System and Security > System > Advanced system settings > Performance > Settings

Then select "Adjust for best performance"

I think I ended up with a pretty basic grey looking display in Reaper, but at least it allowed me to record with lower latency. Then afterwards I could set it back to the setting that says "Let Windows choose what's best for my computer"

Have done this, but appreciate you mentioning it anyway!

thegummy

Stuckfast said:

That's a bit of a weird test in my opinion. What's particularly odd is that they don't state whether they did the latency tests on Windows or Mac OS. I presume it was the latter since the Apogee doesn't have Windows drivers, in which case the latency results are irrelevant to your needs as a Windows user.

The key point here is that Mac OS has a built-in USB Core Audio driver which is actually quite good and makes life easy for developers, so many audio interfaces use that on Mac OS (including the Audiofuse, the JoeCo Cello, Antelope's USB interfaces and, I think, the Audient interfaces). Most of those developers also use a generic third-party driver on Windows, usually the Thesycon one. Unfortunately, this is crap, or at least it always used to be (I haven't tested a Windows interface for a while as I no longer have an up-to-date machine) and delivers poor low-latency performance on the same interfaces that do quite well on Macs.

If you have a fast enough computer and you run at a high sample rate you can get the measured round-trip latency to look OK on most interfaces -- but what that doesn't tell you is how much of a CPU hit you'll take in doing so. The beauty of RME's drivers is that they offer stable low-latency performance without loading down the CPU too much. Having said that, the newer Focusrite USB driver is also very good.

If you want proper evaluation of low-latency performance on Windows you are much better off looking at Vin Curigliano's DAWbench stats.

This is something I've been trying to research recently - would the same PC set to the same buffer size or for roughly the same latency allow more plugin instances before glitching using a Babyface Pro vs. a Focusrite Scarlett?

I haven't been able to find any dawbench results that show both those on the same PC.

Essentially, at the moment I have a Scarlett and let's say I have a project where adding an additional reverb plugin causes glitches, pops or clicks (and removing the plugin fixes it), if I swapped the Scarlett for a Babyface Pro would it allow that extra plugin or maybe even several more?

Branshen

thegummy said:

Stuckfast said:

That's a bit of a weird test in my opinion. What's particularly odd is that they don't state whether they did the latency tests on Windows or Mac OS. I presume it was the latter since the Apogee doesn't have Windows drivers, in which case the latency results are irrelevant to your needs as a Windows user.

The key point here is that Mac OS has a built-in USB Core Audio driver which is actually quite good and makes life easy for developers, so many audio interfaces use that on Mac OS (including the Audiofuse, the JoeCo Cello, Antelope's USB interfaces and, I think, the Audient interfaces). Most of those developers also use a generic third-party driver on Windows, usually the Thesycon one. Unfortunately, this is crap, or at least it always used to be (I haven't tested a Windows interface for a while as I no longer have an up-to-date machine) and delivers poor low-latency performance on the same interfaces that do quite well on Macs.

If you have a fast enough computer and you run at a high sample rate you can get the measured round-trip latency to look OK on most interfaces -- but what that doesn't tell you is how much of a CPU hit you'll take in doing so. The beauty of RME's drivers is that they offer stable low-latency performance without loading down the CPU too much. Having said that, the newer Focusrite USB driver is also very good.

If you want proper evaluation of low-latency performance on Windows you are much better off looking at Vin Curigliano's DAWbench stats.

This is something I've been trying to research recently - would the same PC set to the same buffer size or for roughly the same latency allow more plugin instances before glitching using a Babyface Pro vs. a Focusrite Scarlett?

I haven't been able to find any dawbench results that show both those on the same PC.

Essentially, at the moment I have a Scarlett and let's say I have a project where adding an additional reverb plugin causes glitches, pops or clicks (and removing the plugin fixes it), if I swapped the Scarlett for a Babyface Pro would it allow that extra plugin or maybe even several more?

I can do a couple of tests over the weekend and report back. I'll run both interfaces on some heavy projects with the same buffer settings and see whether the babyface can cope better or if they perform the same. 

thegummy

Branshen said:

thegummy said:

Stuckfast said:

That's a bit of a weird test in my opinion. What's particularly odd is that they don't state whether they did the latency tests on Windows or Mac OS. I presume it was the latter since the Apogee doesn't have Windows drivers, in which case the latency results are irrelevant to your needs as a Windows user.

The key point here is that Mac OS has a built-in USB Core Audio driver which is actually quite good and makes life easy for developers, so many audio interfaces use that on Mac OS (including the Audiofuse, the JoeCo Cello, Antelope's USB interfaces and, I think, the Audient interfaces). Most of those developers also use a generic third-party driver on Windows, usually the Thesycon one. Unfortunately, this is crap, or at least it always used to be (I haven't tested a Windows interface for a while as I no longer have an up-to-date machine) and delivers poor low-latency performance on the same interfaces that do quite well on Macs.

If you have a fast enough computer and you run at a high sample rate you can get the measured round-trip latency to look OK on most interfaces -- but what that doesn't tell you is how much of a CPU hit you'll take in doing so. The beauty of RME's drivers is that they offer stable low-latency performance without loading down the CPU too much. Having said that, the newer Focusrite USB driver is also very good.

If you want proper evaluation of low-latency performance on Windows you are much better off looking at Vin Curigliano's DAWbench stats.

This is something I've been trying to research recently - would the same PC set to the same buffer size or for roughly the same latency allow more plugin instances before glitching using a Babyface Pro vs. a Focusrite Scarlett?

I haven't been able to find any dawbench results that show both those on the same PC.

Essentially, at the moment I have a Scarlett and let's say I have a project where adding an additional reverb plugin causes glitches, pops or clicks (and removing the plugin fixes it), if I swapped the Scarlett for a Babyface Pro would it allow that extra plugin or maybe even several more?

I can do a couple of tests over the weekend and report back. I'll run both interfaces on some heavy projects with the same buffer settings and see whether the babyface can cope better or if they perform the same. 

That would be very kind of you if you have the time!

Branshen

thegummy said:

Branshen said:

thegummy said:

This is something I've been trying to research recently - would the same PC set to the same buffer size or for roughly the same latency allow more plugin instances before glitching using a Babyface Pro vs. a Focusrite Scarlett?

I haven't been able to find any dawbench results that show both those on the same PC.

Essentially, at the moment I have a Scarlett and let's say I have a project where adding an additional reverb plugin causes glitches, pops or clicks (and removing the plugin fixes it), if I swapped the Scarlett for a Babyface Pro would it allow that extra plugin or maybe even several more?

I can do a couple of tests over the weekend and report back. I'll run both interfaces on some heavy projects with the same buffer settings and see whether the babyface can cope better or if they perform the same. 

That would be very kind of you if you have the time!

From my very unscientific testing of running both interfaces at the same buffer rate, it appears that they both are able to avoid glitches, pops and clicks if they are set to a decent buffer size. For the project I was using, this was 128 samples. buffers of less than 64 samples crackled constantly for both interfaces. So, the interfaces appear to perform the same at the same buffer settings.

As I understand it, the reason that audio crackles is because the CPU load is too high, i.e. the CPU cannot keep up with the amount of audio processing that needs to be done. I understand that the main reason to lower buffer sizes is to reduce latency. However, the CPU load will increase with lower buffer sizes, as the CPU essentially has to process data quicker. so one way to avoid crackles is to upgrade to a better CPU which will allow you to use lower buffer sizes without crackling (i.e. your computer is processing the audio fast enough)

If reducing latency is the ultimate goal, upgrading to the babyface did reduce the latency significantly, whilst maintaining the buffer size. the babyface pro can reach sub-10ms at 128 samples, but the scarlett can't.

Branshen said:

Had a good go tonight. First thing I noticed was that it is alot better than my old 2i4 which struggled to play large projects without crackles and constant dropouts. (I had to devise a workaround and export a master track everytime I loaded a session for mixing to stop the crackles) 

I realise that I may have given misleading information here. I probably could have avoided the problem of crackling when I was using the scarlett if I had just set my buffer size high enough. Having a higher buffer size would likely to have allowed the scarlett to work as well as the babyface. At the same buffer settings, the scarlett would have higher latency, but this is a non-issue as latency is not an issue when mixing.

stratman3142

Branshen said:

...If reducing latency is the ultimate goal, upgrading to the babyface did reduce the latency significantly, whilst maintaining the buffer size. the babyface pro can reach sub-10ms at 128 samples, but the scarlett can't...

Just curious but what roundtrip (input/output) latency did you get from the Scarlett at the following spls settings :-
64 spls
96spls (if that's an option)
128 spls

I ask because I'm thinking of the new Solo 3rd Gen model as a possible future option to use with my laptop as a second portable system, after I've double checked what it can cope with to avoid pops and crackles. 

Branshen

@stratman3142 ;;;

I use a windows 10 i7 laptop, with my 1st gen 2i4, 44.1kHz & latest focusrite asio drivers, I get

32 samples - 10.5ms rountrip

64 samples - 13.3ms roundtrip

128 samples - 18.1ms roundtrip

96 samples is not an option.

With my babyface pro, I get

48 samples - 4.69ms rountrip (no typo here, 48 samples gives me higher latency than 64 samples, no idea why)

64 samples - 4.33ms roundtrip

96 samples - 6.41ms roundtrip

128 samples - 7.23ms roundtrip

stratman3142

Branshen said:

@stratman3142 ;;;;;;;

I use a windows 10 i7 laptop, with my 1st gen 2i4, 44.1kHz & latest focusrite asio drivers, I get

32 samples - 10.5ms rountrip

64 samples - 13.3ms roundtrip

128 samples - 18.1ms roundtrip

96 samples is not an option.

With my babyface pro, I get

48 samples - 4.69ms rountrip (no typo here, 48 samples gives me higher latency than 64 samples, no idea why)

64 samples - 4.33ms roundtrip

96 samples - 6.41ms roundtrip

128 samples - 7.23ms roundtrip

Many thanks. I didn't realise it was a 1st gen. Sorry if you mentioned that earlier. I believe they greatly improved the latency when they introduced the Scarlett 2nd gen and I think the 3rd gen is supposed to be at least as good as the 2nd gen.

I thought my Zoom UAC-2 was good, but your babyface pro is better and about 1ms less on those settings.

Branshen

stratman3142 said:

Branshen said:

@stratman3142 ;;;;

I use a windows 10 i7 laptop, with my 1st gen 2i4, 44.1kHz & latest focusrite asio drivers, I get

32 samples - 10.5ms rountrip

64 samples - 13.3ms roundtrip

128 samples - 18.1ms roundtrip

96 samples is not an option.

With my babyface pro, I get

48 samples - 4.69ms rountrip (no typo here, 48 samples gives me higher latency than 64 samples, no idea why)

64 samples - 4.33ms roundtrip

96 samples - 6.41ms roundtrip

128 samples - 7.23ms roundtrip

Many thanks. I didn't realise it was a 1st gen. Sorry if you mentioned that earlier. I believe they greatly improved the latency when they introduced the Scarlett 2nd gen and I think the 3rd gen. is supposed to be at least as good as the 2nd gen.

I thought my Zoom UAC-2 was good, but your babyface pro is about 1ms less on those settings.

Yes, I've read that the latency of the 2nd and 3rd gen scarletts are much better than the first gen.

I was very interested in the Zoom (particularly the UAC-8), but ultimately the, reported stability, hardware monitoring and expandability of the RME won me over. Plus I got a very good deal and managed to get a babyface pro for less than the Zoom UAC-8.

thegummy

Branshen said:

thegummy said:

Branshen said:

thegummy said:

This is something I've been trying to research recently - would the same PC set to the same buffer size or for roughly the same latency allow more plugin instances before glitching using a Babyface Pro vs. a Focusrite Scarlett?

I haven't been able to find any dawbench results that show both those on the same PC.

Essentially, at the moment I have a Scarlett and let's say I have a project where adding an additional reverb plugin causes glitches, pops or clicks (and removing the plugin fixes it), if I swapped the Scarlett for a Babyface Pro would it allow that extra plugin or maybe even several more?

I can do a couple of tests over the weekend and report back. I'll run both interfaces on some heavy projects with the same buffer settings and see whether the babyface can cope better or if they perform the same. 

That would be very kind of you if you have the time!

From my very unscientific testing of running both interfaces at the same buffer rate, it appears that they both are able to avoid glitches, pops and clicks if they are set to a decent buffer size. For the project I was using, this was 128 samples. buffers of less than 64 samples crackled constantly for both interfaces. So, the interfaces appear to perform the same at the same buffer settings.

As I understand it, the reason that audio crackles is because the CPU load is too high, i.e. the CPU cannot keep up with the amount of audio processing that needs to be done. I understand that the main reason to lower buffer sizes is to reduce latency. However, the CPU load will increase with lower buffer sizes, as the CPU essentially has to process data quicker. so one way to avoid crackles is to upgrade to a better CPU which will allow you to use lower buffer sizes without crackling (i.e. your computer is processing the audio fast enough)

If reducing latency is the ultimate goal, upgrading to the babyface did reduce the latency significantly, whilst maintaining the buffer size. the babyface pro can reach sub-10ms at 128 samples, but the scarlett can't.

Branshen said:

Had a good go tonight. First thing I noticed was that it is alot better than my old 2i4 which struggled to play large projects without crackles and constant dropouts. (I had to devise a workaround and export a master track everytime I loaded a session for mixing to stop the crackles) 

I realise that I may have given misleading information here. I probably could have avoided the problem of crackling when I was using the scarlett if I had just set my buffer size high enough. Having a higher buffer size would likely to have allowed the scarlett to work as well as the babyface. At the same buffer settings, the scarlett would have higher latency, but this is a non-issue as latency is not an issue when mixing.

Really appreciate it.

It's very interesting that they perform similarly at the same buffer settings. The buffer setting is arbitrary, what we really want is an acceptable latency. So given that the RME has lower latency at the same buffer settings, it means that if we ignore the buffer number and set both interfaces to target < 6ms latency then the RME will actually allow more plugins to be run on the same system.

The decision would be easy if it wasn't for how big the price difference is, the RME really is very expensive.

It's good to know that it's an option though, if I get this new powerful system and there's still issues getting < 6ms latency (I've found that's what I need to feel real time) with a full project then getting the RME interface would give another step forward.

thegummy

stratman3142 said:

Branshen said:

@stratman3142 ;;;;;;;

I use a windows 10 i7 laptop, with my 1st gen 2i4, 44.1kHz & latest focusrite asio drivers, I get

32 samples - 10.5ms rountrip

64 samples - 13.3ms roundtrip

128 samples - 18.1ms roundtrip

96 samples is not an option.

With my babyface pro, I get

48 samples - 4.69ms rountrip (no typo here, 48 samples gives me higher latency than 64 samples, no idea why)

64 samples - 4.33ms roundtrip

96 samples - 6.41ms roundtrip

128 samples - 7.23ms roundtrip

Many thanks. I didn't realise it was a 1st gen. Sorry if you mentioned that earlier. I believe they greatly improved the latency when they introduced the Scarlett 2nd gen and I think the 3rd gen is supposed to be at least as good as the 2nd gen.

I thought my Zoom UAC-2 was good, but your babyface pro is better and about 1ms less on those settings.

Here are the figures for the Scarlett gen 2:

16: 4.0ms

32: 4.8ms

64: 8.2ms

128: 13.0ms

Do you have the figures for the Zoom? I keep reading it's very good for latency - if there was a reasonably priced interface that had the high performance of the RME without the other features that bump the price up to 600 quid I'd be well in to that.