Hi! I am wondering if I could get some pointers on how to better use A&H ZEDi-10 as an USB interface with a Linux-based DAW for live sound. My current problem is a noticeable albeit small latency between the plucked guitar string and the sound from the monitors. I suppose it should be fixable because if it is noticeable for a beginner like me, it would be unusable for professionals.
I am a beginner, and I have finally gotten around to trying my newly acquired Allen & Heath ZED i-10FX mixer as an interface. It can send 4 independent channels via USB. I installed Ubuntu Studio distribution, with low-latency kernel, and trying with Ardour 6, which comes with it.
The interface work out of the box with this setup but I hear that delay. I know next to nothing about DAWs yet, I was barely able to figure out how to configure the Ardour session to receive and send the sound back. So, perhaps I need to adjust some parameters from defaults.
I use ALSA as sound engine/bus but I also tried JACK with no difference in the delay. Googling suggested that for simple use like mine - playing a single guitar through - there is no advantage of JACK over ALSA, and Ardour itself recommends ALSA.
I tried to calibrate the latency by plugging the monitor out into the guitar in and running the calibration in ardour - it worked, and I got 81 but I am not yet sure what it means, whether these are even milliseconds or frames or something else.
I am also using a pretty powerful i7 quad core 12th generation laptop, so I should not have a bottleneck there.
Any suggestions? Thanks!
I am a beginner, and I have finally gotten around to trying my newly acquired Allen & Heath ZED i-10FX mixer as an interface. It can send 4 independent channels via USB. I installed Ubuntu Studio distribution, with low-latency kernel, and trying with Ardour 6, which comes with it.
The interface work out of the box with this setup but I hear that delay. I know next to nothing about DAWs yet, I was barely able to figure out how to configure the Ardour session to receive and send the sound back. So, perhaps I need to adjust some parameters from defaults.
I use ALSA as sound engine/bus but I also tried JACK with no difference in the delay. Googling suggested that for simple use like mine - playing a single guitar through - there is no advantage of JACK over ALSA, and Ardour itself recommends ALSA.
I tried to calibrate the latency by plugging the monitor out into the guitar in and running the calibration in ardour - it worked, and I got 81 but I am not yet sure what it means, whether these are even milliseconds or frames or something else.
I am also using a pretty powerful i7 quad core 12th generation laptop, so I should not have a bottleneck there.
Any suggestions? Thanks!
DAWs are always going to have some latency.
Generally it can be adjusted by reducing buffer sizes. But doing this increases the risk of audio dropouts due to buffer underruns. Whether you can get the latency low enough to not be disturbing to you is a personal thing: some are fine with 10mSec+ latency, some hate it. As a rule of thumb... sound travels around one foot per millisecond, so a 10mSec latency is about the equivalent of being 10 feet away from the sound source (guitar amplifier, in my case).
When recording on my DAW, I monitor the parts I am recording completely in the analogue domain - i.e., the monitored signal path is not via the DAW at all.
I dont know enough about ALSA / JACK / Linux to advise on tweaking / tuning it for latency. But for reference: my DAW input / output latencies are 6.9mSec (96kHz sample rate); that's on an ancient quad core Core i7 desktop, and running Windows 7 with an M-Audio interface.
Cheers, and regards,
Ant.
Generally it can be adjusted by reducing buffer sizes. But doing this increases the risk of audio dropouts due to buffer underruns. Whether you can get the latency low enough to not be disturbing to you is a personal thing: some are fine with 10mSec+ latency, some hate it. As a rule of thumb... sound travels around one foot per millisecond, so a 10mSec latency is about the equivalent of being 10 feet away from the sound source (guitar amplifier, in my case).
When recording on my DAW, I monitor the parts I am recording completely in the analogue domain - i.e., the monitored signal path is not via the DAW at all.
I dont know enough about ALSA / JACK / Linux to advise on tweaking / tuning it for latency. But for reference: my DAW input / output latencies are 6.9mSec (96kHz sample rate); that's on an ancient quad core Core i7 desktop, and running Windows 7 with an M-Audio interface.
Cheers, and regards,
Ant.
Thanks, @Ant Moore ! I think my latencies are of much larger scale, not 10ms. It's not like I maybe hear something comparable with me stepping away from the speaker. It is very clearly "I pluck the strings" - "some time later I hear the sound from the speaker" kind of thing, time delay comparable with the fast chord changes. That is what makes me think I did not configure something right - given that my hardware is fairly modern, and I am using the distribution specifically made for this. Thanks for the hint about buffer size - I will look for that buffer.
I'd comment why would you do it any other way? The board is a mixer, that happens to be able to send 4 channels up into the computer for recording. Use it as such; your amplified speakers connected to the "main mix" outputs, no computer necessary at the initial setup.
Then when everything's tweaked; level, pan, eq, effects and your mix sounds good when you play, then think about recording tracks.
When recording on my mixer, I lose the lush reverb I like while playing through the board to stereo speakers playing off the "main mix" outputs. I guess you're supposed to put that back as an effect from the DAW's arsenal, so you're not stuck with whatever the board's effects were set to.
Trouble is, much of the time that reverb setting is used as part of my performance; it sounds strange on playback without it.
Agreed.
Not how I record. I tend to record a couple of parts at a time.
If you just used the DAW as a tape recorder to record everything in one go, then you;'d need to do some of that. But that defeats the object of using a DAW really...
I record with flat EQ, and as little "comfort" effects as I can get away with. Maybe a dash of analogue compression on the feed to the DAW to prevent hard clipping in the ADC.
I want the recorded tone "right", sure. But level, pan, eq, effects are far better applied in the DAW during mixdown, after the tracks have been recorded. Otherwise, you are stuck with the effects etc you recorded. If there's too much of them, you cannot take them away in the DAW; the only option there is to re-record the offending part(s). Or as I do nowadays: capture the signal directly off the guitar lead at the same time as recording the mic'ed up amp. If I need to re-record any parts for any reason, I can just feed the DI'ed signal back to the amp and record that.
It is a problem, yes.
If you must have them when tracking, then so be it. Though the DI trick can help.... If you capture a DI of your perfect performance as well as the mic'ed up version, then if at mixdown those effects are too much you can re-track it from the DI with less effects applied.
Cheers, and regards,
Ant.
The answer is when you are paid for a session ( in professional context) you are paid for what is present on tape/ HDD not what is sent to the recording device. So during a professional session monitoring is always done on playback returns not at desk input or recorder send: this allow to be sure you didn't miss anything happening within the recording device and printed to tape/hdd.
Of course you are at home so you do what please you but here is answer to your initial question.
Regarding latency, it's not the first time i hear about very high latency through linux... i wonder if it's not an issue in the OS ( kernel or drivers are not optimised enough stock?). Under windows i never had such high latency and software monitoring under native system ( non dsp based) never been an issue since 2002. I commonly reduce buffer size when tracking ( especially when there is few instruments playing together -overdub situation- the exception being drums of course... ).
Within my home system i'm under 3ms overall (in to out) hardware latency ( RME aes soundcard).
About effects, it's always nice to have an hardware reverb and small desk to treat things for monitoring purpose. The latency compensation implemented into DAW often create more issue than they solve in tracking.
I always track a straigth DI of any instrument as well as mic or 'regular' output: it gives an 'escape way' ( save your .ss) in case something was wrong with mic or hardware choice... reamping capability is nice option to have ( even if non mandatory). For bass DI ( straight, no fx) are mandatory! 😉
Regarding Linux latency, I don't know first hand yet but I see many reports over at least last decade saying that it is used professionally. So I assume it has no critical flaws. My setup is not tuned in any way yet but it is a distribution that is packaged for this purpose, with special low latency kernel.
I'm guessing that my problem is a newbie operator error...
I'm guessing that my problem is a newbie operator error...
Could you point me to the professional use of Linux you have seen Crocobar?
I know some very powerful tools are based on it in hifi but not i know of in proworld per se. The simple reason about it is there is almost zero driver support for major brands of easily availlable recognised in pro world soundcards.
That said i'm not located in same side of the world and habits can be different ( eg: Cubase is the 'standard' DAW for semipro in here EU, Logic was more common in US... the defacto standard in pro studio always been Apple/Protools everywhere in the world).
But as it is a very open OS maybe it is possible to optimize further this issue, idk? ( hence my previous comment).
Anyway, yes if you have not optimised buffer size it's sure it is not top performance yet.
I know some very powerful tools are based on it in hifi but not i know of in proworld per se. The simple reason about it is there is almost zero driver support for major brands of easily availlable recognised in pro world soundcards.
That said i'm not located in same side of the world and habits can be different ( eg: Cubase is the 'standard' DAW for semipro in here EU, Logic was more common in US... the defacto standard in pro studio always been Apple/Protools everywhere in the world).
But as it is a very open OS maybe it is possible to optimize further this issue, idk? ( hence my previous comment).
Anyway, yes if you have not optimised buffer size it's sure it is not top performance yet.
Last edited:
@krivium I couldn't but I am not one myself, and a newbie to top - what do I know? But I've seen testimonials, even on this site I think. Also, perhaps it depends on what you consider professional use? I imagine that studios wouldn't use Linux as a rule because the expense is trivial to them, and they will go with the most common solution I imagine. But I would consider many musicians who are not stars with piles of money but are excellent musicians, who record their own stuff, to be professional enough.
I would consider them to be professional enough too.
The issue is this: most semi-pro / excellent musicians are just that: they are not generally software geeks. And for all the advances in linux over the recent years, it is still a labour of love to get things optimised to run well with a pro / semi-pro grade audio interface even if linux drivers exist for it. So your semi-pro excellent musician is more likely to get a suitable audio interface for an existing laptop (Windows or Apple), and some semi-pro DAW software to suit. A number of audio interfaces ship with stripped-down DAW software that is good enough to get going.
To be fair: I don't know much about the quality / feature sets / usability of DAW software in the linux world. I've personally never heard of. or witnessed anyone using a linux system for such a purpose.
Cheers, and regards,
Ant.
Fair enough, perhaps I'll be able to share my experience in a while. I figured that if there is a whole OS distribution dedicated to this, and numerous DAWs like Reaper and Ardour and LMMS, someone's making them for some reason.
I suspect it's like in everything else: most people don't use Linux at home, it would be alien and disorienting for them. But some - like myself - do it exclusively their whole life, and there is nothing more natural. My whole family, including children on their PCs use Linux exclusively (lately I had to use MAC for work but that is pretty isolated to work). I imagine that I am in a relatively small minority but probably not negligible.
Both of my kids manage their rather extensive gaming setups in Linux, and I imagine that DAW stuff is not much more complicated, nor the musicians less capable, at least some of them.
But I really don't know, just guessing.
Last edited:
Hi,
Thank you for your answer Crocobar.
Professional to me means literraly to earn money from an occupation. In that case then i've not seen anybody using Linux based system in studio. Maybe for composing but not from the client i have/had in the past few years ( even if some of them still compose with Atari!).
It's not an issue with what the OS can do, it really goes down to 2 issues: first one is hardware related ( as already pointed point me to Linux drivers for the major names in the industry. Motu might be the exception i know of but they are usually seen as ( very good but) semipro gear).
Second one is compatibility. One of the major concern we have in studio is related to exchange of projects between facilities. It's not uncommon to have multiple studios involved into mix and record of one project and there is a need for compability between them. Hence why ProTools on Mac was the defacto standard ( mainly because the cards used in Digidesign systems were developed for mac...).
That said you don't care about that, you are at home doing things for fun. I was just curious about what you said and if i missed a trend.
Have you played with buffer size since then?
Thank you for your answer Crocobar.
Professional to me means literraly to earn money from an occupation. In that case then i've not seen anybody using Linux based system in studio. Maybe for composing but not from the client i have/had in the past few years ( even if some of them still compose with Atari!).
It's not an issue with what the OS can do, it really goes down to 2 issues: first one is hardware related ( as already pointed point me to Linux drivers for the major names in the industry. Motu might be the exception i know of but they are usually seen as ( very good but) semipro gear).
Second one is compatibility. One of the major concern we have in studio is related to exchange of projects between facilities. It's not uncommon to have multiple studios involved into mix and record of one project and there is a need for compability between them. Hence why ProTools on Mac was the defacto standard ( mainly because the cards used in Digidesign systems were developed for mac...).
That said you don't care about that, you are at home doing things for fun. I was just curious about what you said and if i missed a trend.
Have you played with buffer size since then?
I'm just guessing too, but if you are hearing obvious latency it might be a USB Streaming mode issue:
Safe = 16 ms
Extra Safe = 32 ms
That's getting near slapback echo territory, 31ms is a 1/64th note at 120 BPM.
http://www.allen-heath.com/media/ZEDi-USB-Windows-Driver-Help-Manual-1.pdf
Hope the problem fix is that simple!
Hope it is too! With any luck, the linux driver for the desk has the same settings / options.
FWIW, an activate-able W10 isnt that hard to get. They have scripts that allow a fresh install to bypass most of the crap load that ordinarily comes with it. One would think to the point where it just runs the program you want - nothing else happening - particularly if you never connect it to the internet after the initial activation.
I'm too lazy to try all that, even though I read it can be done. Even too lazy to turn off automatic updates on the machine I run my DAW on; they try to sell me something every time I turn it on after their latest, even when disconnected from the 'net.
I'll end here by guessing a properly setup W10 system for DAW use would be on par with wrestling with all the drivers and other idiosyncrasies a Linux system would present. You're going to have to go through something, in order to get it to work right.
I'm too lazy to try all that, even though I read it can be done. Even too lazy to turn off automatic updates on the machine I run my DAW on; they try to sell me something every time I turn it on after their latest, even when disconnected from the 'net.
I'll end here by guessing a properly setup W10 system for DAW use would be on par with wrestling with all the drivers and other idiosyncrasies a Linux system would present. You're going to have to go through something, in order to get it to work right.
I did, last weekend but did not have enough time to really get an idea. I found the setting for the bufer size, and I think the smaller buffer size made it sound near-perfect but I got distracted by something else, will have to do a better test in the next few days.
I did more experimentation with the Ardour 6 running within the Ubuntu Studio Linux install, and the buffer size 128 is the lowest I was able to get to work well. The latency is either not there or small enough to not bother me so far. Interestingly, stepping another factor of two down to buffer size 64 turns the sound into a completely distorted garble. I wonder if I could upgrade the driver or something to push it further, or this is a hard limit for my setup.
Its not surprising, the lower the buffer size, the more often the DAW / driver needs to load data into the buffer to avoid buffer underrun.
The more the DAW does, the more likely it is to run out of steam and underrun.
I tend to run the buffer as small as I can during tracking, if I'm monitoring via the DAW. But during the mixdown phase, when latency is less important, then depending on the complexity of the project I will increase the buffer size rather than run on the edge of underrun.
Cheers, and regards,
Ant
The more the DAW does, the more likely it is to run out of steam and underrun.
I tend to run the buffer as small as I can during tracking, if I'm monitoring via the DAW. But during the mixdown phase, when latency is less important, then depending on the complexity of the project I will increase the buffer size rather than run on the edge of underrun.
Cheers, and regards,
Ant