Lessons learned from researching Audio Interfaces

Discuss music production with Ableton Live.
mikb
Posts: 373
Joined: Sat Jan 08, 2011 9:53 pm

Re: Lessons learned from researching Audio Interfaces

Post by mikb » Sat Dec 09, 2017 12:05 am

login wrote:I can get my fireface UC in my i7 desktop down to 32 samples, I can load a couple of instruments and effects before getting CPU overload.

I think at the end all RME USB interfaces use the same drivers, so same performance. Sample rate, bit depth, channel count and your own CPU and project size is what makes the difference. But RME can go as low as 32.
It can, but look at those latencies at 32 and 64. They are the same.
Last edited by mikb on Sat Dec 09, 2017 12:08 am, edited 1 time in total.
Basic gear info: Macbook Pro with macOS 10.12, Ableton Live Suite version 9 (64bit) with Ozone, Push and APC20 as controllers.

fishmonkey
Posts: 4478
Joined: Wed Oct 24, 2007 4:50 am

Re: Lessons learned from researching Audio Interfaces

Post by fishmonkey » Sat Dec 09, 2017 12:38 am

a couple of thoughts:

- input and output latencies are not the same, and can be affected differently by changes in buffer size, so you need to consider both measurements separately.
- i don't understand why 8.5 ms roundtrip is "unacceptable"? that's the latency of an acoustic instrument being played about 3 metres away from you.

Mister Natural
Posts: 275
Joined: Wed Mar 21, 2007 2:44 pm
Location: michigan

Re: Lessons learned from researching Audio Interfaces

Post by Mister Natural » Sat Dec 09, 2017 5:52 am

cmprvndncr wrote:
patrickstinson wrote:...and then back to the K-Mix outputs (at 96 kHz, with a buffer of 64 samples)."[/i] so that seems totally unnaceptable... With my 2i4 I'm looking at just under 8ms round-trip latency (reported in live at least)...cause this system will be used for a perform-everyday corporate gig on a ship in a few months
here's the thing I don't understand about your post - why the 96khs setting ? If you're on stage, in a club or whatever, no one in the audience(or even likely yourself) will notice the slightest difference between music through loudspeakers being processed at 44.1 or 48 or 96. I think that at least some of the latency bottleneck comes from this high sample rate. This oversampling will also eat up a lot more hard-drive storage space. Totally unnecessary for your application.
https://en.wikipedia.org/wiki/Sampling_ ... rocessing)

best of luck
an expert only on what it feels like to be me
& you are who you google
#smile

fishmonkey
Posts: 4478
Joined: Wed Oct 24, 2007 4:50 am

Re: Lessons learned from researching Audio Interfaces

Post by fishmonkey » Sat Dec 09, 2017 6:06 am

Mister Natural wrote:
cmprvndncr wrote:
patrickstinson wrote:...and then back to the K-Mix outputs (at 96 kHz, with a buffer of 64 samples)."[/i] so that seems totally unnaceptable... With my 2i4 I'm looking at just under 8ms round-trip latency (reported in live at least)...cause this system will be used for a perform-everyday corporate gig on a ship in a few months
here's the thing I don't understand about your post - why the 96khs setting ? If you're on stage, in a club or whatever, no one in the audience(or even likely yourself) will notice the slightest difference between music through loudspeakers being processed at 44.1 or 48 or 96. I think that at least some of the latency bottleneck comes from this high sample rate. This oversampling will also eat up a lot more hard-drive storage space. Totally unnecessary for your application.
https://en.wikipedia.org/wiki/Sampling_ ... rocessing)

best of luck
as long as your computer has the power to do the calculations in time, increasing the sample rate lowers the latency, as each sample is of shorter duration.

for example, at the same buffer size, doubling the sample rate halves the buffer duration, and potentially halves the latency...

cmprvndncr
Posts: 126
Joined: Wed Mar 01, 2017 2:44 pm

Re: Lessons learned from researching Audio Interfaces

Post by cmprvndncr » Sat Dec 23, 2017 8:39 am

Hi guys, thankyou for all your input on this. It's helped me get closer to buying an interface.

Apologies that I haven't been on here for a bit, my family has been going through some dramas, but back in the saddle now!

I managed to get to my local rme reseller, who are really helpful w/ some of the best customer serviceive experienced in the world (store DJ in Sydney) (guitar factory, Parramatta also up there). Anyways, they had a ucx which was out of its box, so I was able to fire it up on my laptop (2014 15" mbp, 16gb ram, SSD, 2.5ghz CPU)... So at minimum buffer size, the rme is rock solid, running a moderately busy session... No clicks or pops. Those drivers really are amazing, as compared to my 2i4, which just can't hack it at 32, even with minimal plugins.

So I'm ready to spring for an rme, the problem is that I'm getting a MUCH higher latency than others reported here... I went back the next day, as I thought maybe I had lookahead limiters and the like adding extra latency, and fired up a blank session, but the latency seemed much the same...

So I'm clocking at:

@ 44.1 - 1.54ms input & 2.11ms ouput latency, RTL: 3.65 (almost double your report latency!)
@ 96k - 0.9ms input & 1.16ms output latency, RTL: 2.05... was hoping for less than this at 44.1 :'(

Am I doing something wrong? Is there a latency setting in the totalmix software or something?
patrickstinson wrote: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.
Patrick, can I ask what puter your using?

As far as which RME interface suits me best, the babyface pro looks most ideal... I really don't need the extra I/O of the UCX, but I'm already not happy with the reported latency in live on the UCX, the babyface is allegedly slightly higher in latency...

So, a bit confused at the moment, but I need a solution real soon, so any further advice is really preciated.

Here's some pics of the buffer / RTL info as reported above - https://imgur.com/w1MhEPH https://imgur.com/SlKMxiN

scheffkoch
Posts: 593
Joined: Thu Jan 08, 2015 12:37 pm

Re: Lessons learned from researching Audio Interfaces

Post by scheffkoch » Sat Dec 23, 2017 10:08 am

...without having read all of your comments: why do you think 3.65 ms is a "high" latency?...
regarding rme totalmix: this could give you the option to have a "zero latency" setup...by listening through totalmix (and not the daw) you'll get the most minimal technical possible latency...but beware of possible pdc problems...
macbook pro m1pro, macos monterey, rme multiface via sonnet echo express se I, push 2, faderfox mx12, xone:k2

cmprvndncr
Posts: 126
Joined: Wed Mar 01, 2017 2:44 pm

Re: Lessons learned from researching Audio Interfaces

Post by cmprvndncr » Sat Dec 23, 2017 10:21 am

scheffkoch wrote:...without having read all of your comments: why do you think 3.65 ms is a "high" latency?...
regarding rme totalmix: this could give you the option to have a "zero latency" setup...by listening through totalmix (and not the daw) you'll get the most minimal technical possible latency...but beware of possible pdc problems...
Hey there, thanks for chiming in there... 3.65 is relatively low, but it's higher than the OP, much higher. So I want to try to get to the bottom of why... Is it simply my macbook model, or something I can address without needing a new expensive computer (I just bought this one a month ago.)

scheffkoch
Posts: 593
Joined: Thu Jan 08, 2015 12:37 pm

Re: Lessons learned from researching Audio Interfaces

Post by scheffkoch » Sat Dec 23, 2017 10:51 am

...tbh i don't trust these numbers...i have a late 2016 macbook pro and with my rme multiface 1 (more than 10 years old) i am able to achieve a roundtrip latency of around 8.x ms at 128 samples buffer...since i change instrument presets during live performance (which can lead to cpu/ram buffer spikes) i now tend to work with 256 samples buffer which gives me around 12.5 ms roundtrip latency...since playing software instruments is only affected by the output latency i play at around 6.x ms which is absolutely ok...and for my bandmate that i record while we're playing the 12.5 ms is also absolutely acceptable...actually i didn't tell him that i raised the buffer and he hasn't complained yet...ha...
macbook pro m1pro, macos monterey, rme multiface via sonnet echo express se I, push 2, faderfox mx12, xone:k2

cmprvndncr
Posts: 126
Joined: Wed Mar 01, 2017 2:44 pm

Re: Lessons learned from researching Audio Interfaces

Post by cmprvndncr » Sat Dec 23, 2017 11:39 am

scheffkoch wrote:...tbh i don't trust these numbers...i have a late 2016 macbook pro and with my rme multiface 1 (more than 10 years old) i am able to achieve a roundtrip latency of around 8.x ms at 128 samples buffer...since i change instrument presets during live performance (which can lead to cpu/ram buffer spikes) i now tend to work with 256 samples buffer which gives me around 12.5 ms roundtrip latency...since playing software instruments is only affected by the output latency i play at around 6.x ms which is absolutely ok...and for my bandmate that i record while we're playing the 12.5 ms is also absolutely acceptable...actually i didn't tell him that i raised the buffer and he hasn't complained yet...ha...
Thanks for sharing your experience with it. For me, if I was playing along with other people, or even to track, my latency needs would be much less demanding, but similarly to the OP, I'm doing live looping with the Looper plugin, so every ms counts.

cmprvndncr
Posts: 126
Joined: Wed Mar 01, 2017 2:44 pm

Re: Lessons learned from researching Audio Interfaces

Post by cmprvndncr » Sat Dec 23, 2017 11:42 am

PS. When you say you don't trust those numbers, do you think they are over or under the true amount? Also, how do you check? I've done the built-in lesson on setting the buffer compensation setting thing nudet the latency info in preferences, where you bounce from one track to another, but it came back sayung that the numbers are on point. Do you use a different method?

scheffkoch
Posts: 593
Joined: Thu Jan 08, 2015 12:37 pm

Re: Lessons learned from researching Audio Interfaces

Post by scheffkoch » Sat Dec 23, 2017 2:59 pm

...you do know that the buffer compensation setting is only relevant when you listen directly through your interface and not the software? this is because when you listen to the "dry" signal you play your instrument "dry", whereas when you listen through the daw your ears and your brain automatically compensate the "delay" because your ears will hear the sound including the latency reported by the daw and thus the software can easily compensate it (because it knows the value)...for 5 five years i "only" did live looping with audio (i used mainstage and the mobius plugin) but latency was never a problem for me as long as the "delay" was under/around 10ms...i must say i learned to play piano but i'm far from a professional piano player who may have a problem with such latency values, e.g. when playing very fast staccatos...to summarize: when you're comfortable with your daw's latency, meaning when you can play your instruments without the feeling of "always being too late", your live looping will be on point...
macbook pro m1pro, macos monterey, rme multiface via sonnet echo express se I, push 2, faderfox mx12, xone:k2

cmprvndncr
Posts: 126
Joined: Wed Mar 01, 2017 2:44 pm

Re: Lessons learned from researching Audio Interfaces

Post by cmprvndncr » Sat Dec 23, 2017 3:39 pm

scheffkoch wrote:...you do know that the buffer compensation setting is only relevant when you listen directly through your interface and not the software? this is because when you listen to the "dry" signal you play your instrument "dry", whereas when you listen through the daw your ears and your brain automatically compensate the "delay" because your ears will hear the sound including the latency reported by the daw and thus the software can easily compensate it (because it knows the value)...for 5 five years i "only" did live looping with audio (i used mainstage and the mobius plugin) but latency was never a problem for me as long as the "delay" was under/around 10ms...i must say i learned to play piano but i'm far from a professional piano player who may have a problem with such latency values, e.g. when playing very fast staccatos...to summarize: when you're comfortable with your daw's latency, meaning when you can play your instruments without the feeling of "always being too late", your live looping will be on point...
Thanks again for your insight... As I understood, the buffer compensation settings were related to the internal bus system being aligned a sample accurate level, but maybe I had it all wrong?

It's interesting that the OP said he haid issues w/ that kind of latency... I've been using hardware loop pedals (mostly rang III w/ sidecar), and so my muscle memory is really trained for whatever super low latency that pedal has... But I can learn on a new system. Your post makes me feel like my reported latency might be ok... However, I'm still anxious to learn why I'm gettig such dramatically different results to the OP...

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

Re: Lessons learned from researching Audio Interfaces

Post by Khazul » Sun Dec 24, 2017 7:30 pm

If you want really low latency for live use - you need to go up to 96K and have a beefy CPU. In which case I suggest pretty much anything recent from RME as an interface.
Nothing to see here - move along!

[jur]
Site Admin
Posts: 5307
Joined: Mon Jun 01, 2015 3:04 pm
Location: Ableton

Re: Lessons learned from researching Audio Interfaces

Post by [jur] » Sun Dec 24, 2017 9:57 pm

3ms is absolutely nothing. Do you worry because of the number (compared to what others may have reported) or to your feeling while playing? If you can feel it while playing then you're probably from an other universe, and then I doubt anyone here can help you about how-to outer-universe music playing.
But if you figure it out at some point don't forget to make a YouTube tutorial.
Cheers
:P
Ableton Forum Moderator

scheffkoch
Posts: 593
Joined: Thu Jan 08, 2015 12:37 pm

Re: Lessons learned from researching Audio Interfaces

Post by scheffkoch » Mon Dec 25, 2017 12:58 am

…here are some interesting real world numbers…:
https://www.gearslutz.com/board/music-c ... -base.html
macbook pro m1pro, macos monterey, rme multiface via sonnet echo express se I, push 2, faderfox mx12, xone:k2

Post Reply