Direct monitoring via interface

jimmys69

jimmys69

MOODerator
I am curious about how and why it is recommended for those new to recording to use the 'Direct monitoring' ability of an audio interface. I personally do not understand how that even works. Stay with me...

So, if you are recording a track using direct monitoring from any interface, are you still not having to perform to the DAW's output? I assume that gets rid of any latency while recording, but at the expense of not being able to have effects in monitoring without DSP ability or just hearing the mix from DAW output?

Granted, my system is not a typical off the shelf PC, and my interfaces likely have better performance than the lower priced stuff, but is that really that big of a difference between a interface with higher latency drivers and the system it is running on? I have never had such an issue myself though when I decided to do this at home (after experience in other digital based studios), I went with a more powerful system than some would likely go with. So I have a gap in what I am able to suggest to those starting out.

I consistently record vocals on fully mixed 60+ audio and 5 or so VSTi's running projects with buffer set at 96 samples with 5.96 ms of latency and no dropouts. There are VST's like FabFilter Pro L that create enough latency due to it's over-sample processing that creates an unusable delay while recording. But I just haven't wrapped my head around why I would use the direct monitoring of recordings.

Is this just based upon system abilities and or the interface and it's driver abilities to record at low latency? I feel I may have missed a step by going big at first and wish to know how to help others that haven't.
 
I'm in the camp to not use any effects when recording vocals. Let the singer hear what he sounds like without masking anything. Reverb can hide pitch variations. Compression can hide dynamics and vocal control. These are things a singer needs realtime feedback to be able to deliver a good performance. Newb singers need a crutch to get started and reverb in the cans provide them that comfort. They really need to step up and learn to sing without the gimmicks. Then add the gimmicks during mixdown.

With direct monitoring, you're still hearing the output from the playback channel of the DAW, it's blended in with the input channel that you're recording. Most interfaces give you the ability to control the balance of playback to recorded channel so the performer gets a good balance. Oddly enough, my UAD Apollo does it in a convoluted way and it's the only thing about it I don't like.

When I first started, I used a firewire mixer. It was a great little device. It would stream raw audio to the computer, but I could add reverb to the monitor of the channel I was recording without it going to the computer. Reverb in the cans, yeah man. Later I switched to a mic pre and standalone converter arrangement. I couldn't get effects in my headphones, so I sang without. I think it made a big difference for me. I learned to hear myself and my pitch control improved over time.

So that's my take. Direct monitoring is good. And you don't have to worry about latency.

(Of course, with the UAD Apollo, I can record with all kinds of effects, none going to tape, and not have any latency. I use a channel strip. No reverb)
 
Well, I can't answer for every system or DAW, but...

With my Audition and ASIO based interface, when I go to Direct Monitoring, the drivers manage the timing of the tracks so that playback and new stuff via the mic(s) remain in sync. A few years ago I did a test and the sync accuracy was pretty well spot on...less than 1ms difference.

I can get pretty low latency monitoring with the round trip via the computer but it's never quite as good as direct--and the latency varies with the buffer size I have to set which in turn goes up as I add more tracks and effects on what's already recorded.

I should add that adding effects to the headphone monitor mix (but still recording dry) is not an issue for me--I can record via a digital mixer with built in effects, and set up all sorts of custom monitor mixes while still recording dry.
 
I too have never really paid much attention to that...mainly because I track to tape before dumping to DAW, so I'm always monitoring the direct signals, and latency doesn't even enter the equation...but that said, there are times when I have recorded straight to the DAW, and my interface's virtual mixing console allows me to direct monitor...I don't even do anything to make that happen, it's just how it works. It's like a straight foldback off the interface.
Now...how do the already recorded tracks work with that...?
Well...again, in my case, even when I track to the DAW, my old-school, analog tape mentality treats the DAW like I would a tape deck...so, I don't ever start adding FX or what have you to those tracks...therefore, I think that's why even with the recorded tracks...I'm not hearing any latency, and I can set my buffers pretty low, so it doesn't come into play.

I think (as Bobbsy has also said) that the interfaces are creating some "digital adjustment" to the pre-recorded tracks to keep them in sync with the new/direct tracks.
Of course...not having dealt with any latency issues, I've never had the need to investigate that stuff further.

I'm sure lots of folks will record a track, then drop a bunch of processing and FX on it...then go record another track...etc...etc...and that puts a much heavier load on the system doing all that real-time processing, keeping the pre-recorded tracks and new/direct track in sync...and to make it work, buffers get set larger, and now the latency starts to become noticeable.

I don't say this for any "analog VS digital" reasons...but having seen lots of folks dealing with serious latency issues, I've often smiled at the fact that because I still track to tape before going to DAW, I have ZERO issues during tracking.
Plus, I can also add outboard processing/FX to the tape tracks just for my cue mixes, and still no latency to deal with.
Then, when I'm in the DAW...all the tracks have been transferred, so again, I don't deal with latency. :)
 
I am also in that camp to some degree though I am also working with a few artists that are used to being 'in the moment' for lack of working phrase. The ones that are not just starting out and have years of experience singing and recording do actually sing better in 'context' of the final mix sound. This seems to work for them best because they are looking to have the sound they envision for the final mix. Not so much for the few I have in here that are not so experienced.

In any case, Reverb is the lowest on my list of tracking effects that I use. More would be the two level compression I have on the vocal group the tames the vocal levels a bit but more for the quality of the API 2500 that I find useful pretty much always. In my ears, it does not taint the performance by leveling it out, it to me shows what the end result will be. If it doesn't work, then I would know to do another take because there was obviously something wrong.

But that doesn't often happen with the seasoned singers I have been blessed to work with. There are times here I will even use something kind of extreme like a Waves Doubler while tracking a rap line to place them in the mode that they are looking to achieve. To me, there are times when recording vocals with 'un-printed' effects is just like playing a guitar part with a wah. You can't play the same guitar part without the wah under your foot any more than you can be effective singing a part without the desired effect that was decided upon in pre-production or on the fly. I suppose I am sometimes placing the cart before the horse in some ways, but without knowing what the horse can do with the cart? I could be wasting time.

Other techniques may work for others and there is no correct/one way to do anything. I would never disrespect any method.

I am just honestly curious where this changes and where the processing power or ability of an interface makes it a necessity to use direct monitoring. I do have full DSP effects that I could use with direct monitoring, but I have never had need to use them.

Though the foundation of my house is literally falling down, I broke my right hand, lost an employee after he cut of a couple finger tips and his uncle bled to death the next day, and my family almost died from carbon monoxide poisoning in the last few weeks. Maybe I need to change my ways...
 
I mostly use direct monitoring; not because I'm wedded to it, but because it is usually the most convenient and helpful way of recording.

There are times when I sue software monitoring, particularly if an effect is to be an integral part of the sound, and needs to be there to get the treatment right in the first place.
 
I 'm in the 'no effects' on my vocal when recording camp, too, so direct makes most sense. The reason its advised for noobies, is they ARE using cheaper interfaces, don't have their latency in single-digit milliseconds and they're usually complaining about an 'echo'.
 
Yeah, direct monitoring is mostly a solution to what was a real problem for most people even just a few years ago. With a decent interface and more modern computer, though, it shouldn't really be necessary in most cases.

I don't think it's the drivers themselves, but the DAW that does the compensation for I/O latency. It does depend on the driver actually reporting the correct latency to the DAW. Most of the time it just works, but I have heard of some interfaces that somehow don't give the right numbers, and you end up having to do loopback tests to figure it out and override it using the DAW's tools. Once that's set, though, you don't really have to worry about it. Play along to what's coming out of the computer and it'll end up lined up as closely as you played it.

I use one of my machines as a mixer and effect rack for live work all the time. In fact, my entire band goes through the thing for amp sims and drum sounds and vocals with EQ and compression. I also mix other folks through it occasionally and nobody ever complains about latency. Course, this is quad core 3.9GHz, but you don't quite need that much grunt just to play back a few tracks and record a vocal without latency. If you're really strapped for resources, you might need to freeze any tracks with effects or VSTis during tracking.

Of course, you do have to be careful adding plugins with their own latency. Oversampling doesn't really have to introduce latency depending on how it's done, but I guess it can. Shouldn't be more than a few samples (Edit - no, I just can't figure out why OS would add latency at all), but who knows what else those goofy fuckers are doing. I think FFT based EQs and such have to add latency because they have to fill a "window" before they can start to do their thing. Things like pitch shifters will also have to do this. Of course, anything with lookahead pretty much has to add latency as well because you can't actually look ahead when the input is live. In mixing, I like to use a lot of really long "pre-comp" settings in ReaComp, but I can't slap that on during tracking or in a live setting, because it introduces like a quarter second delay. Watch out for gates and things with "pre-open" also. Some of the "fancier" limiters get more smoosh with less distortion by using a bit of lookahead, and some don't actually tell you they're doing it.
 
Last edited:
The new computer at my friend's studio is probably more than fast enough to monitor through the DAW, but I haven't tried it yet. For that matter I don't even use the direct monitoring, I just use the mixer. At my other friend's studio there is no mixer, just a rack of preamps and an Apollo 16. That works okay too, but the Console (direct monitoring control panel) isn't a clear win over a hardware mixer. Latency is lowish but not low enough through his Pro Tools system so we use the direct monitoring.
 
On big projects I use my mixer and set up monitoring through there. On small stuff though, I have a basic 2 channel interface with direct monitoring and that can be pretty useful for setting up a quick headphone mix.

No matter how fast your computer and HDDs there's always a question of "how much latency is too much" and, in my experience, different people have different tolerances. For that matter, it also depends on what instrument is being recorded. I've found that drummers are much more fussy--but I guess that makes good sense. If they WEREN'T sensitive to timing issues, they shouldn't be drummers!

As for effects in the monitoring, I'm always recording other people, not myself so I basically give them what they want. Reverb is the most common request I get so I even have a preset ready to go on my mixer for that. However, I never record the effects--my tracks are always dry. Well, that's not quite true. On certain recordings (usually when working on something live where I don't get a lot of time for level setting and rehearsal) I'll sometimes add a compressor on each track. However, this is set up purely as a limiter and doesn't change the signal unless I get sloppy or the lead singer screams unexpectedly. Just occasionally, it can be a get out of jail free card!
 
Bobbsy said:
No matter how fast your computer and HDDs there's always a question of "how much latency is too much" and, in my experience, different people have different tolerances. For that matter, it also depends on what instrument is being recorded. I've found that drummers are much more fussy--but I guess that makes good sense. If they WEREN'T sensitive to timing issues, they shouldn't be drummers!
Click to expand...

Well I will ask about the 'No matter how fast your computer and HDDs there's always a question of "how much latency is too much" comment'. Around 5.3ms for me typically with my gear @ 96 sample buffer. But it is rare that I have a project large enough with a full mix worth of FX and VSTi tracks that I ever get over 7ms with buffer settings at 128 samples when my system starts to act funky. Performance can start to get erratic as far as audio glitches while moving parts or cutting events while recording, but the audio being recorded is not being interrupted. I myself can't even hear latency below 9ms and I am pretty anal about that. I am not even sure a human ear can sense 6 ms of delay unless they are super sensitive in an autistic way...

Again, I am only curious as to why those that use direct monitoring feel the need and why they still do. I understand the need for some due to poor performance of their systems, but it seems that that is not as much an issue for those who have a modern system.

I do wonder where the need ends in regards to computer power and interface ability. What makes it a necessity to have the knob to mix direct and computer signals if a system is able to give low latency results? - because I personally have not had an experience that warranted that.

In no way am I talking down anyone who works in a different way. I merely posed the question as to why they feel the need to. In a situation where using a mixer for monitoring before tracking then it is obvious why. But then I also am curious the need to do that still.

I am likely rambling. Been a long few weeks...

Because of issues I can't control, I have to rebuild my studio. I am looking into other possible setups while I have the chance to start over.
 
If people can play well with a stage monitor 5 or 6 feet from their ears, and a whole bunch of other arrival times from other monitors, amps, drums etc. then 5 or 6 milliseconds of latency is generally no big deal. But there are also situations in which 1ms makes an audible difference.

The mixer setup I use is a legacy from the days of DA-38s. I swapped out the Tascam tape decks for PCI based MOTU interfaces and just kept on recording, mixing ITB instead of on the M-2600.
 
My DM1000 is specced at (from memory) a maximum latency of 4ms (input to output including effects, less with no effects) and I've never had anyone notice it.

However, I suspect that Jimmy's system is a much higher spec that a lot of people who post in here and that the "HR average" would be rather more latency that he's getting. It can pretty quickly become a problem with a typical non-specialist system with (say) a 5400 rpm HDD and the need for relatively large buffers. Being newbieist here, it can also be easier to tell a beginner how to set up monitoring if it's just a turn of a knob on an interface rather than trying to work out how to route input back to output on a variety of DAWs. It's probably me just being lazy that I always recommend Direct monitoring when people ask for interface suggestions.

Of course, then we get into MIDI and you have to have low latency monitoring on the round trip!
 
Ah! MIDI! My son never had a problem with latency (2496 card, Cubase Elements 6) playing a keyboard, even if he was writing drum parts.
The snag was playing TO parts already in the PC. The fastest combination of the sound card and Cubase was only just good enough to keep replay and live play in sync.

Yes of course, musicians can an do learn to live with huge delays (why the LSO etc have a conductor!) but we is not talking about people with that kind of skill or training. (justfort! We talk of a "tight" rhythm section, especially in jazz. Could this be because the players tend to crowd each other?)

Dave.
 
Hmmm...thinking out loud here:

While we talk about a "tight" group of musicians, we also complain about how things like an electronic drum kit fail, often saying "they're too perfect".

Could it be because we're used to the very small differences in sound caused by the spacing between musicians...and even between individual drums? After all, the space between the bottom of the bass drum and top of the cymbals is probably 3 or 4 ms...and the vocalist could easily be 10ms in front of the drums.
 
Bobbsy said:
Hmmm...thinking out loud here:

While we talk about a "tight" group of musicians, we also complain about how things like an electronic drum kit fail, often saying "they're too perfect".

Could it be because we're used to the very small differences in sound caused by the spacing between musicians...and even between individual drums? After all, the space between the bottom of the bass drum and top of the cymbals is probably 3 or 4 ms...and the vocalist could easily be 10ms in front of the drums.
Click to expand...

I'll take "tight/perfect" over all that "space/loose" stuff because many people who object to tight/perfect do so because they have a hard time achieving it. Oh, and tight.perfect doesn't mean every note of every track is lined up perfectly on some timing grid...meanwhile, space/loose often means one guy is ahead while another is behind, etc...which doesn't really sound all that good.

Each song is different, and while there may be an overall groove for a given song that sounds right when the bass is a little behind the kick, or when the rhythm guitar leads the beat...etc...that still has to remain tight/perfect throughout the song for the groove to sound good throughout the song.

When it comes to drums, IMO, they, being the reference time keeper, need to be tight/perfect, which then allows the other instruments to set the rest of the groove, and find that feel a little ahead, right on, a little behind...etc.
The "bad" thing with a lot of sampled drum tracks that people churn out has to do with poor sequencing...too much on the grid...to much of the same rat-tat-tat sounding velocities...or some attempt to create a loose groove without finding what actually sonds good for that song...etc....which can all go to making some sampled drum tracks sound fake.

I think that gets compound by the fact that many musicians these days are doing most if not all of their tracks solo...so they are playing them one at a time...and it's not easy for them to find the right groove for each instrument. IOW...they may nail the bass groove, but then lose sight of where the rhythm guitar should be, because they're just doing each track as some standalone thing...same way people will record one track and then immediately start applying processing and FX to it wihtout considering the bigger mix picture.

Creating a cohesive mix isn't something you only do when you get to the mixing stage...it has to start during tracking...ESPECIALLY if you are a solo musician doing one track at a time. You have to really pay extra attention to that groove building as you add track after track...whereas a room full of musicians can find it all together at one time.
 
jimmys69 said:
I am curious about how and why it is recommended for those new to recording to use the 'Direct monitoring' ability of an audio interface. I personally do not understand how that even works. Stay with me...

So, if you are recording a track using direct monitoring from any interface, are you still not having to perform to the DAW's output? I assume that gets rid of any latency while recording, but at the expense of not being able to have effects in monitoring without DSP ability or just hearing the mix from DAW output?

Granted, my system is not a typical off the shelf PC, and my interfaces likely have better performance than the lower priced stuff, but is that really that big of a difference between a interface with higher latency drivers and the system it is running on? I have never had such an issue myself though when I decided to do this at home (after experience in other digital based studios), I went with a more powerful system than some would likely go with. So I have a gap in what I am able to suggest to those starting out.

I consistently record vocals on fully mixed 60+ audio and 5 or so VSTi's running projects with buffer set at 96 samples with 5.96 ms of latency and no dropouts. There are VST's like FabFilter Pro L that create enough latency due to it's over-sample processing that creates an unusable delay while recording. But I just haven't wrapped my head around why I would use the direct monitoring of recordings.

Is this just based upon system abilities and or the interface and it's driver abilities to record at low latency? I feel I may have missed a step by going big at first and wish to know how to help others that haven't.
Click to expand...

Well the difference between direct monitoring and DAW monitoring is the signal is turned around inside the interface's dsp mixer instead of going in and out of the computer. For example, it takes 45 uS to convert analog to digital. and another 45 uS to convert it back to analog. so direct monitoring in this example is 90 uS. Now if we use DAW monitoring, you have to include the time the interface makes the round trip into and out of the computer. So if the interface takes 10mS to make the round trip the actual latency would be 10.090 mS in this example.

on the question of system abilities, its the real time performance of the system.

attached is a video explaining this:

 
Last edited:
I always monitor direct. It just makes it easier to achieve the performance I'm aiming for if I take latency out of the picture. I'd rather work with my limitations as a musician without the complication of slop in monitoring. If you are adding lots of effects ITB and need to hear those in tracking, I get it. But I don't do that. Apart from adding reverb in the mix (usually only to vocals and drums, and as little as I can get away with), I track everything that will be on the guitars, bass, and keyboards. So there is no downside to direct monitoring for me.
 
You must log in or register to reply here.
Back
Top