Lessons learned from researching Audio Interfaces

Discussion of music production, audio, equipment and any related topics, either with or without Ableton Live
patrickstinson
Posts: 168
Joined: Thu Oct 13, 2011 5:01 am
Location: Anchorage, Alaska
Contact:

Lessons learned from researching Audio Interfaces

Post by patrickstinson » Tue Jan 28, 2014 5:14 pm

I am a live violin / bass / soft synth looper, and spent over two years researching audio interfaces to find the lowest latency possible for a macbook. Sadly if you *really want to know the facts about what’s out there today, you have to dig really really deep into endless forums and technical data, and even still can only infer the information that will actually make an impact on your purchase. Hopefully this will save you some time.

I'm not an expert but did spend a lot of time researching and shopping. Please feel free to tear this post apart if you feel like anything in it isn’t correct.

THE PROBLEM

I have a latest model retina macbook pro with maxed out options, and was using an old Edirol FA-101 over FW400 (FW400 -> FW800 -> TB). The lowest buffer size I could use in Live was 64 samples @ 44100, which gave me about 8-9ms round trip latency.

Despite the word on the street, this latency is not acceptable for live work. My band-mates were constantly complaining about timing. When trying to use Ableton Looper with a MIDI pedal, it is impossible to lock in the loop on time. It’s also not possible to play perfectly in sync with a Ping-Pong delay.

I only need a few channels, but am also willing to pay for quality. The goal is to find the fastest small-format interface possible.

THE STATE OF THE MARKET

Shopping for interfaces is extremely hard right now. Most reviews are several years old (2006-2008?) and on hardware that doesn’t compare with what we have today. For example, it’s extremely difficult to find hard interface latency numbers even for the very popular macbook pro / Ableton combo. Ideally we would have a professional Ableton-centric round-up of interfaces on MacBooks (or PC!).

DRIVERS

First thing: if an interface is going to be fast, it must have optimized drivers from the manufacturer. This basically includes Apogee, LYNX, RME, and some M-Audio interfaces. From tons of forum searching, I became convinced that nobody’s drivers can compare with RME. I found lots of dirt on Apogee, have so-so personal experience with M-Audio, and have found absolutely nothing but praise for RME’s speed and stability. So it seems like the choice is clear.

THE BUS

Next you have to choose a bus. We are currently in a period of minimal innovation for audio interfaces. Firewire is being phased out and USB 3.0 and Thunderbolt are starting to build momentum. USB 2.0 is plenty sufficient for say 16 channels, but USB 3.0 will make more channels available (although at the same latency as USB 2.0).

The quality of your USB chipset will affect your latency. Various tech forums report that lots of older or even current PCs are built with sub-standard chipsets, and there are enough differences to make publishing latency numbers impossible. I don't know what this is like now, but the Apple chipsets are great.

It’s also important to note that Thunderbolt is actually PCIe. Using a TB interface you can expect PCI performance but they usually use adaptors like Apogee Symphony IO + Thunderbridge ($3000 total), or RME HDSP + Sonnet Echo PCI -> TB ($1200 and up). This appears to be the ideal performance situation, but they are expensive and physically clunky.


TAKING THE LEAP

Having a macbook, the choice of USB VS Thunderbolt still remained for me. I considered going with the RME HDSP and Sonnet Echo, but couldn’t for the life of me figure out how to get the usual I/O XLR/TRS connection options found in the cheaper interfaces. So the USB FireFace UC or UCX started looking more attractive.

Plus, RME repeatedly says that their USB products compete with their PCIe products for latency, but there are no hard experiential numbers available online. Also you have to dig deep, but you will find that the RME Fireface UCX has newer AD/DA converters that are about 1ms faster than the Fireface UC, which is a big deal.

But, there is absolutely nowhere that reported on the maximum performance of any USB interface with a current computer, and this would have been the most helpful info to have. Still I took a chance and went with the UCX.

RESULTS - INTERFACE

I was really blown away with the speed and stability of the FireFace UCX over USB. I can now use Looper and play perfectly in sync with Ping-Pong delay. At a buffer size of 32 samples @ 44100, Live is reporting < 2ms round trip, and I believe it. The latency is low enough that I can’t tell the amplification apart from the instrument's natural sound until it is taken away. The TotalMix software is really incredible. The only feature I'm missing is a wifi interface and iPad app to use the UCX as a standalone mixer.

So there you have it. I would love to see others post their latencies with Live on various interface / computer combinations. That’s what I feel I was missing most in my search.

RESULTS - ABLETON LIVE

One important thing to note is that you can't really get a feel for Live's performance ceiling until you have eliminated the audio interface variable. So I will fleetingly say that Live's performance at the basics is pretty damned good. I'm controlling 7 loopers from a midi pedal with a reasonable amount of effects on each channel. I do get some audio scratches when starting Live's transport via the first Looper, but usually only when using a Max For Live device that listens to tempo changes, for example buffer shuffler. This info obviously doesn't cover all cases, but it does show that the boilerplate performance is great and the rest is up to you.

LINKS (please post more)

http://www.dawbench.com/audio-int-lowlatency2.htm
Last edited by patrickstinson on Tue Jan 28, 2014 10:12 pm, edited 2 times in total.
Web | Mobile | Audio
http://vedanamedia.com/

login
Posts: 1816
Joined: Sat Jul 16, 2011 12:41 am

Re: Lessons learned from researching Audio Interfaces

Post by login » Tue Jan 28, 2014 5:58 pm

I agree with everything, except for some reason it doesn't seem to came across infor on Steinberg interfaces which have received quite good feedback around music production forums.

But yeah, if you want the best and never upgrade again (or in many years) RME is the way to go, happy Fireface UC owner here :)

Khazul
Posts: 3185
Joined: Wed Feb 23, 2005 5:19 pm
Location: Reading, UK

Re: Lessons learned from researching Audio Interfaces

Post by Khazul » Tue Jan 28, 2014 7:02 pm

Could have saved you a lot of time - just start with RME, and your choice is how many inputs and output do you want? The actual connection technology just doesn't matter as in the end both windows (ASIO) and OSX demand a certain amount of buffering. Windows can potentially be a tiny bit better on a fast desktop machine with well chosen parts (48 samples achievable).

If you want the lowest possible latency and can afford to suitably over-spec you CPU, then run at 96K (or even 192K) sample rate as much lower latency is achievable, but obviously it needs vastly higher CPU use to achieve it.

I use a UFX over USB on a Feb 2011 MBP - it can manage 64 samples at 44.1 even within a decent size project (I normally use 128 or 256 for project work). I have even had it doing 64 sample at 192K before which Live indicates as 1.14ms - thats less than the typical delay though a hardware midi rack synth when using an external midi controller/sequencer and for just using the laptop purely as an fx box or synth from an external controller playing live - the cpu use is OK. Obviously its not OK when trying to work on a big project.

The other thing I will say for RME - they seem to provide driver support for a long time. The end result might be that a high end RME interface over say 10 years might work out cheaper than the 3 other similar feature audio interfaces you might have had to buy due to shitty companies who are in the habit of screwing their customer base over driver support at every opportunity. After being badly screwed in the past by this - it was one of my main motivators for spending a fortune on a UFX knowing that the end result would be that I would have solid drivers very good audio hardware (mic-preamps, DACs, ADC etc) and low latency and probably get driver support for 10 years+ after buying it.
Nothing to see here - move along!

leisuremuffin
Posts: 4721
Joined: Tue Apr 06, 2004 12:45 am
Location: New Jersey
Contact:

Re: Lessons learned from researching Audio Interfaces

Post by leisuremuffin » Tue Jan 28, 2014 9:22 pm

i run at 88.4k specifically to achieve less latency while not burning as much cpu as 96k or higher would.

my macbook is about 4 years old or so.

i use a motu ultralight and easily achieve 4.6ms total latency for when i am recording live instruments or using live as my guitar stompbox and looper. i bump the buffer up to 512 for around 13ms total latency when i'm doing studio work or playing live sets to give me more room on my cpu.
TimeableFloat ???S?e?n?d?I?n?f?o

oblique strategies
Posts: 3577
Joined: Thu Nov 02, 2006 9:57 pm
Location: Another Green World

Re: Lessons learned from researching Audio Interfaces

Post by oblique strategies » Tue Jan 28, 2014 11:10 pm

Thanks for the report.

RME Fireface UC here. TotalMix is essential.

turnitto11
Posts: 60
Joined: Fri Jun 22, 2012 6:35 am

Re: Lessons learned from researching Audio Interfaces

Post by turnitto11 » Wed Jan 29, 2014 2:36 am

Most people have latency issues because they want to run all these channels through a DAW and monitor through their DAW. That is the WRONG way to record live audio. You either need to have a hardware mixer with direct outputs to send to an audio interface into your DAW, so you can monitor through the mixer and have 0 latency, OR you must have an interface with software/driver that allows you to mix monitors before it hits the the analog to digital conversion.

People seem to bitch about the PreSonus FireStudio stuff for latency all the time, but frankly I think people are just using it wrong. It's a cheap interface, sure, but it's all about how you use it. I set my buffer on Live very high (1024 for the best recording stability), turn the monitor within Live OFF, and then I mix all my monitor's through PreSonus' Universal Control software/driver. Universal Control is essentially a digital control for a hardware mixer. You can mix the direct signal being sent to the interface straight to the outputs of the interface. So, you have audio entering the interface, and the signal gets sent straight to the output for monitoring, as well as a signal getting sent through FireWire into Live and you hard drive. I have literally had 0 latency issues since discovering the magic and power of Universal Control because there is NO latency. And you can still mix in audio from Live/DAW with your direct stuff, so you can play along with clicks and playback audio, no problem.

And lots of interfaces today come with software that let you have this direct mixing control, but people usually just disregard them because they seem confusing. Like I said, I haven't had latency issues since I found out how to use these resources. Before, yes, I ran through Live and monitored through Live, and I was battling latency all day. Those days are over.
MacBook Pro (13", 2.7ghz i7, 16gb, 10.9), Suite 9

fishmonkey
Posts: 4088
Joined: Wed Oct 24, 2007 4:50 am

Re: Lessons learned from researching Audio Interfaces

Post by fishmonkey » Wed Jan 29, 2014 3:44 am

turnitto11 wrote:Most people have latency issues because they want to run all these channels through a DAW and monitor through their DAW. That is the WRONG way to record live audio. You either need to have a hardware mixer with direct outputs to send to an audio interface into your DAW, so you can monitor through the mixer and have 0 latency, OR you must have an interface with software/driver that allows you to mix monitors before it hits the the analog to digital conversion.
no, monitoring through your DAW is not "the wrong way".

direct monitoring is the only way to solve the latency problem if your interface and computer combo are too slow to do the required processing on the fly. and of course you sacrifice the ability to use software effects in your signal chain.

processing power has progressed to the point where a good setup can be used on the fly, as other people have mentioned above.

this is not just about recording live audio either, it's also an issue for people who perform live. if you are using software effects, then of course you would rather hear the effected sound as you play, not a direct monitored dry version.

aisling
Posts: 2640
Joined: Fri Mar 10, 2006 8:58 am
Location: 50 miles north of SF

Re: Lessons learned from researching Audio Interfaces

Post by aisling » Wed Jan 29, 2014 4:45 am

I am closely following how the new Motu 828x will perform. I am in the market to upgrade, and am a happy motu user. I am waiting for the reviews to pour in...
http://www.motu.com/newsitems/introduci ... technology
http://soundcloud.com/aislingbeing


Live, Reason, Moog sub phatty, Moog sub 37, Ozone 6, guitars, Pedals, proper ergonomic sitting posture, french pressed coffee with a pinch of cardamon.

Colorshape
Posts: 3
Joined: Fri Feb 12, 2010 9:43 pm

Re: Lessons learned from researching Audio Interfaces

Post by Colorshape » Wed Jan 29, 2014 9:59 am

Live was 64 samples @ 44100, which gave me about 8-9ms round trip latency.
At a buffer size of 32 samples @ 44100, Live is reporting < 2ms round trip
... may be i'm dumb, but i don't quite get that.

May be you should'nt trust too much 'reported latencies, but actually 'measure them ?

patrickstinson
Posts: 168
Joined: Thu Oct 13, 2011 5:01 am
Location: Anchorage, Alaska
Contact:

Re: Lessons learned from researching Audio Interfaces

Post by patrickstinson » Wed Jan 29, 2014 7:04 pm

... may be i'm dumb, but i don't quite get that.

May be you should'nt trust too much 'reported latencies, but actually 'measure them ?
The first numbers are with my old FA-101, the second are with the new RME Fireface UCX. I should have made clear that these were the lowest buffer settings I could use for clean audio with my Live set for each controller.
Web | Mobile | Audio
http://vedanamedia.com/

tecgen
Posts: 46
Joined: Wed Mar 03, 2010 4:14 pm
Location: Germany
Contact:

Re: Lessons learned from researching Audio Interfaces

Post by tecgen » Thu Jan 30, 2014 7:53 pm

patrickstinson wrote:The only feature I'm missing is a wifi interface and iPad app to use the UCX as a standalone mixer.
There is an TotalMix FX app for iPad available on the AppStore. I bought it, but couldn't find time to try it out.

Here's the link to the RME forum with some screenshots.
http://www.rme-audio.de/forum/viewtopic.php?id=18787

best regards,
Marco

patrickstinson
Posts: 168
Joined: Thu Oct 13, 2011 5:01 am
Location: Anchorage, Alaska
Contact:

Re: Lessons learned from researching Audio Interfaces

Post by patrickstinson » Thu Jan 30, 2014 8:18 pm

tecgen wrote:
patrickstinson wrote:The only feature I'm missing is a wifi interface and iPad app to use the UCX as a standalone mixer.
There is an TotalMix FX app for iPad available on the AppStore. I bought it, but couldn't find time to try it out.
It is important to note that the TotalMix FX iPad app only works when connecting the Fireface directly to the iPad in Class-Compliant mode. So this is not a wifi remote control app when using the FireFace in standalone mode, which would effectively give you a great digital mixer. I am aware that other interfaces provide this function, for example the Presonus interfaces.
Web | Mobile | Audio
http://vedanamedia.com/

mikb
Posts: 373
Joined: Sat Jan 08, 2011 9:53 pm

Re: Lessons learned from researching Audio Interfaces

Post by mikb » Mon Jun 23, 2014 12:10 am

turnitto11 wrote:Most people have latency issues because they want to run all these channels through a DAW and monitor through their DAW. That is the WRONG way to record live audio.

Before, yes, I ran through Live and monitored through Live, and I was battling latency all day. Those days are over.
For those of us using mainly software instruments latency is still a big issue. There's no real alternative to monitor trough Live then, nor is it possible to use a larger buffer when recording.
Basic gear info: Macbook Pro with macOS 10.12, Ableton Live Suite version 9 (64bit) with Ozone, Push and APC20 as controllers.

slicedbread
Posts: 313
Joined: Tue Sep 12, 2006 5:22 pm

Re: Lessons learned from researching Audio Interfaces

Post by slicedbread » Tue Jun 24, 2014 10:50 pm

i've been researching a bit for a step up from my dinky all-in-one xiosynth and i think i've settled on the roland quad capture. all the reviews are good on the preamps, headroom and reportedly solid drivers. not too expensive either.

Harmonic Progression
Posts: 84
Joined: Sun Jun 15, 2014 1:06 am

Re: Lessons learned from researching Audio Interfaces

Post by Harmonic Progression » Wed Jun 25, 2014 2:22 am

turnitto11 wrote:...People seem to bitch about the PreSonus FireStudio stuff for latency all the time, but frankly I think people are just using it wrong...

...I set my buffer on Live very high (1024 for the best recording stability), turn the monitor within Live OFF, and then I mix all my monitor's through PreSonus' Universal Control software/driver. Universal Control is essentially a digital control for a hardware mixer. You can mix the direct signal being sent to the interface straight to the outputs of the interface. So, you have audio entering the interface, and the signal gets sent straight to the output for monitoring, as well as a signal getting sent through FireWire into Live and you hard drive.

...

And lots of interfaces today come with software that let you have this direct mixing control, but people usually just disregard them because they seem confusing. Like I said, I haven't had latency issues since I found out how to use these resources. Before, yes, I ran through Live and monitored through Live, and I was battling latency all day. Those days are over.
I have a Firestudio Project and your technique is intriguing, but it's on a high level and I could really use a bit more detail. Can you direct me to a step-by-step instruction for doing what you are doing?
Desktop: Intel i7 hex-core, 64 GB ram, several SSDs, Focusrite Liquid 56, UAD-2.
Laptop: Alienware Area 51: Intel i9, 32 GB RAM, several more SSDs, UAD Apollo Twin.
Software: Win 10, Cubase, Live+Push2, Maschine, and more.

Post Reply