Hear Digital Console Time Issues (Public)

[@Edwin-van-der-Putten]

Wow, Dave... it does give you something to think about if you have the Yamaha mixer and you are in a complicated situation... an accident can easily happen. Once again... you have delivered a great job! I am a big fan of your channel!

[@DaveRat]

Thank you so much Edwin!!

[@tomas.lindroos]

I noticed on my x32r that if you plug in "insert effects" on a group you actually get a bit of delay. I was about to mix a rock gig and wanted to run all the drums, including a semi-big 80:es style reverb into a stereo group, and then use an insert compressor on the group to glue things together. That sounded fine in the PA, but the band also used the P16m personal monitor mixers, and wanted kick and snare dry on separate channels, and the rest of the drums on a stereo channel (the whole drums subgroup). We then got phase issues in the monitor mixes between kick, snare, and the drums bus. Maybe the kick and snare outputs to the P16m could have some delay added to them to get them back in phase - I didn't have time to debug this at the gig, so I just sent the overheads in stereo instead of the drums bus... it would have been cool to get some of the FX and such sent to the monitor mix also. I must take a new look at this at some point... This proved your point on the need to know about possible gremlins in your console before you show up at a gig and get surprised.

[@DaveRat]

Interesting and even if there is delay to try and compensate it never quite works well because the increments are always too coarse to align properly. Also it gets messy as when you add delay there is a minimal delay of 0.3 milliseconds so you would need to add that on to both and then try and fix. Running audio down multipaths in digital that are later resummed is rarely a good idea

[@platemod8465]

Dave - great illustration once again. No matter the technology, there's always a compromise, which is seldom discussed (seems like it was discussed latency more when dsp for system processing was new). Recently I made a change in an IEM personal mix rig in a house system, where before, the drummer was provided a drum mix and two discreet (direct out) channels. I replaced the 'mix' channel in the IEM's with direct fed overheads, and it changed how the drummers played, for the better. Hindsight now says I'm going to have to review the delta-T for the mix vs the directs on the console. Mike

[@DaveRat]

Cool cool and thank you Mike!

[@NealMiskinMusic]

My strategy for this is parallelization. If you want a parallel compression bus, use two busses and only compress one of them so that both busses get the same amount of latency and there's no time differential between them. Want parallel compression on you lead vocal or maybe your snare mic? Assign the same preamp input to two channels, if your snare is channel 3 assign channels 3 and 4 to input 3 and compress channel 4, rather than routing channel 3 internally to compress it. It sucks to have to use some busses for no reason other than that the console doesn't have good latency compensation, but if the console doesn't automatically compensate for it, then you've just got to make sure that anything that is getting split and summed back with itself goes through processes that induce the same amount of latency on both signal paths.

[@NealMiskinMusic]

And feed in-ears direct from the input channels, I always try to minimize bussing in monitor world. It's worth noting though that a singer standing in front of a wedge is going experience about 4 to 6 milliseconds of delay between the speaker moving and the sound wave hitting their ears, most consoles are in the 3 to 5ms range so it's more delay than the natural sound of one's voice but less than what you'd get from a wedge, and wedges don't seem to bother too many singers, at least when they're properly EQ'd.

[@DaveRat]

Very cool and yes, that's a good strategy. As far as in ear latency and it's audibility, the sealing off of the ears such that one hears more of those internal voice sound add a factor. Actually Hear In-Ear Headphone Latency Issues (Public) kzbin.info/www/bejne/d4HNYY2Xgt5njtk

[@MrJFBro]

So good.

[@DaveRat]

Awesome and thank you!!

[@enifyako]

I discovered this myself when trying to, for ambitious reasons, set up a parallel set of processing on a CL5. There was no way to guarantee synchronous timing without using the exact same signal path length, and trying to get around that was only going to introduce comb filtering. So you either had the full chain, two chains using up similar processing, or no chain at all, and any trying to skirt around the processing latency would give you issues in all sorts of areas related to the console routing, be it picking off monitoring points or side keying, what have you. Sometimes the effect slots would even have this issue on the mix parameter.

[@DaveRat]

👍🤙👍

[@pauluzz1991]

Awesome video Dave, but when you add some time delay on a channel, does the console show the actual time delay or does the latency gets add up to it.. really curious if it does or not. Love your content, from Paul

[@DaveRat]

great question. there are three things. The latency of the console which typically is not shown because it impacts everything going through the console, the added latency caused by inserting or creating the delay which typically is shown as a minimum amount of delay. So like when you engage the delay on an X32 it start at I believe .3 ms. and third there is the delay that is being added.

[@SydHeresy]

Dude! I love your videos! You really help bring the science back into sound engineering. There's so many videos out there that are just subjective points of view, while you put it out EXACTLY the way it is. Love your scientific and semantic videos!

[@DaveRat]

Thank you Syd!!

[@studiodespenates]

Things are getting philosophical in here...for the better ;-) Very good point. Thanks for sharing this.

[@DaveRat]

👍🔧👍

[@Kaxlon]

This is a great topic! Thanks for sharing your knowledge, Dave. For me this video is very trippy. Me, a middle aged HiFi nörd listening to a world class sound engineer from across the glode playing with his multi-$$$$$ consoles sampled by another high-end device. Sound is packed, stored, streamed, and unpacked at my end on a PC which optically feeds a $$$-ish DAC, then via XLR to a Behringer DCX2496, and into 3 amplifiers which powers 6 speaker elements, and finally through my ears to my brain. Then Dave adds relativity into the mix... Damn. Deep. We humans are great inventors. Greetings from Sweden! =)

[@DaveRat]

Thank you Kaxlon very cool and love Sweden! Cant wait to be back there!!

[@ToddWCorey1]

Dave, you are a brilliant man!

[@johnwilliamson467]

Time is just a measurement . The latency added to a signal with out latency is the distortion . I understand the concept you working with here . You do a fine job working with the problem that are faced in the live venue . Very nice video .

[@DaveRat]

Thank You John!

[@iamjohnnytrox]

Slightly off topic, but I've noticed that when you link channels (that are set identical and hard panned) on the m/x32, i think i noticed a slight loss in volume and/or change in tone. Do you have any idea what that would be? Or maybe you could run em both through your oscilloscope in a future video and show the difference. Anyways, love the videos. Keep em coming!

[@DaveRat]

I can look into it next time I set up the x32.. There are several types of common panning methods that have differing strategies ranging from panned center at normal volume and hard panned having a boost to compensate for it being on just one side in an attempt to keep the perceived level consistent. Or having hard panned be normal level and a drop of 6, 3db or 4.5db when panned center depending on whether they panning is based constant power, linear or something between linear and constant power. Accordingly, an input may drop in level, be the same or be increased in level when compared to the same input that has been hard panned or to two inputs that are hard paned. I will ponder a demo vid on the subject.

[@il_moe]

Always pushing it, Dave. Love it. Does inserting rack effects and channel processing add even more?

[@DaveRat]

Yes! This is before adding anything and both analog and digital inserts increase latency further

[@bangopek]

Nice

[@DaveRat]

🤙🤙🤙

[@doougle]

I'd say shift in the output latency is not a distortion because, as you said, the output waveform is only shifted in time but isn't changed. A summation of two different latencies is a distortion because the waveform is changed. It's interesting to ponder such things.

[@DaveRat]

🤙🤙🤙 the fact that the latency can't be undone and that the built in time delays in the console are not precise enough to correct and realign the signal, make latency unwieldy at best.

[@lamixxman]

The internal fx processor inserted on a buss or channel ads an additional amount of latency, on the x32. An audible amount, if double bussing.

[@DaveRat]

Agreed

[@gaffster787]

Many digital signal operations done on noise (such as convolution and correlation) are done by sliding one sample set against another by one sample clock at a time and then doing computations. For example, you can pick out a periodic signal buried in noise with "autocorrelation". At t=0,0 a signal perfectly correlates with itself, however at t=0,1...not so much. As you go on, periodic patterns will begin to emerge (correlate) and the "random" noise falls out. I think latency issues would fall under the category of phase distortion. Group delay (by frequency band) is a huge issue for wideband RF circuitry.

[@DaveRat]

Excellent perspective!

[@HazeAnderson]

I definitely notice undesired phase shifting with my effects pedals that I use with my mixer via its Aux send and returns. I just remove the signal from channel and let the effects pedal process the whole thing, but I would much rather blend it back in without the phase issue. Oh well, at the end of the day processing the entire signal thru effects is not a problem. :)

[@DaveRat]

I could see that, it gets messy with little time delays being added in by all sorts of gear

[@peniku8]

I had this exact discussion with an engineer from Meyer Sound, who set up the PA at the venue and we came with a Soundcraft VI3000 for the main act and an X32 for the support band. He ranted about how the Soundcraft introduces comb filtering when you try to run parallel busses like in the video, while the X32 didn't. I really wonder how DiGiCo's S line compares, generally. It's affordable but I barely see these anywhere.

[@steakikan]

If S line is anything like the AH Dlive counterpart it is mostly latency compensated.

[@DaveRat]

I will test a small digico at some point when our SD11 comes back off tour.

[@DaveRat]

I would think so and it would be interesting to test.

[@peniku8]

@@DaveRat Very cool, looking forward to it! The SD11 is a fun little powerhouse.

[@DaveRat]

🤙🤙🤙

[@robertcharbonneau2283]

Said quickly, latency is a very real issue with live monitoring on stage, where in some instances a 1 mms delay between a direct source (let’s say a guitar amp) and the monitors will require twice the power to be adressed. And we are just talking about amplitude here...

[@DaveRat]

1 ms of latency on one of two perfectly combined signals will give a cancellation at around 560hz or so Not much impact on frequencies below 200 hz and comb filtering at higher frequencies. Definitely can be audible and undesirable

[@orlarieper6175]

Difference in Latency in summed signals is some kind of distortion Or signal degradation an is audible and is something to consider, could also happen with effects or in dsp I guess

[@DaveRat]

🤙🤙🤙

[@weareallbeingwatched4602]

And that's why in LDN where I'm system tech for a dub venue, we're running all-analogue. All the way to FOH, monitors and FM transmitter. No latency anywhere.

[@DaveRat]

Good stuff. I do like avoiding a conversion of possible

[@f-unfairstudiotime]

Excellent video, as always! One perspective you skipped or just brushed upon, and that perhaps most definitely confirms your thoughts and classifies a LIVE digital mixing console's latency as "distortion", i.e. an unwanted deviation, is found within the console's ultimate purpose: it's used to mix, blend, effect, process, control and then amplify a LIVE performance, meaning music/sounds happening here & now, live. As such, they exist from the stage or instruments or voices into the air, within the space where they are being made/performed, and being mixed on that console. Therefore, ANY latency will cause what ultimately comes through the PA or amplification system to be "non-identical" in time to the actual sound at the source, time shifted and thus phase shifted and conflicting within that same air and space. And this is the ultimate failure, the Achille's heel of any digital console or processing in a live scenario, as it fails to exactly and unadulteratedly or undistortedly reproduce what it's meant to be mixing and reproducing for amplification purposes. It's a bit of a philosophical stance, and in this respect it doesn't matter that for all practical purposes the amplified sound can easily drown out the original live sound of the performers and instruments, the fact of the matter is that it is delayed, not by choice but by technical default, as soon as it goes through even just the A/D & D/A process, even without additional processing, plugin or bussing latency and console-specific gremlins & design fails (I've been in live situations where we inserted a de-esser or a multiband compressor in a digital console and the processing latency suddenly got so bad the phase issues made it sound like the whole PA was going to implode). Luckily, analog still loves us and our sounds... ;) BTW, I often work in a recording studio fitted with a big Studer Vista digital console - beautiful sounding preamps and converters, even sweet sounding digital eq - but the analog inserts are not latency compensated, so it has forced my workflow to change, when there: if I want to do any analog processing, I only can do it when recording, before the A/D, and if I mix there, I have to use the console's eqs and processors or ITB plugins, if I want to preserve phase coherence and timing between the recorded instruments, sounds and tracks. And I've worked in a smaller studio based around a far cheaper little digital console and an ethernet based headphone distribution system - and a more fine-eared, sensitive singer has said "What's with my headphone sound? My voice sounds weird..." - and there was nothing I could do about it. Now those, too, albeit not linked to the live reproduction of sound for amplification, are at least to me rather major "distortions" not just of the sound, but also of what I was intending to do and of how I like to do it. All of the above in my opinion & experience, of course. Enjoy, AF

[@DaveRat]

Excellent. And yes agreed with the modifier that a context is needed to determine relevance. As far as pure latency, since console latency is so short and equates to a few inches in physical distance, it can be viewed as irrelevant except when signals of differing latency are summed. But yes, a distortion that is relevant in some scenarios.

[@ToddWCorey1]

I would add that even though there is latency added by digital processes, it is still far outweighed by the time difference between what is produced by the PA and what comes off stage naturally, since the PA is closer to the listener than the backline is. Like Dave said, digital latency amounts to inches while backline/PA difference is feet.

[@DaveRat]

🤙🤙🤙

[@ovoono3050]

What is the model and brand of these closed-back headphones he is wearing?

[@DaveRat]

Denon AH-D5000

[@ovoono3050]

@@DaveRat tnx Dave!

[@DaveRat]

Cool cool, yes, they sound great and go super low.

[@obidavekenobe]

This talk reminds me of the HAAS effect. Hah, I was just thinking about sub arrays😃

[@DaveRat]

👍👍👍

[@SyncA81]

To me, on it's own it isn't, but latency can be a building block of distortion. Btw, thx for the vid!

[@DaveRat]

👍👍👍

[@16LiveRecords]

So CL1 is quite not the one for paralel compression group 🤔

[@DaveRat]

Yeah, or at least use a set of groups for compression and another set with compressors set to no ratio. Which is the best way anyway to do it.

[@studiodespenates]

@@DaveRat are you talking about the digital world only (in the particular case of parallel compression) ?

[@DaveRat]

Yeah, identical signal paths should always be used when recombining digital signals. So for parallel compression, a set of groups with active compressors and a set of groups with identical compressors set to a 1:1 ratio for the non compressed version Then combinied signals will be latency correct

[@iamjohnnytrox]

I always wondered why Yamaha made it difficult to use subgroups on their consoles. I would imagine this is one of the reasons why.

[@DaveRat]

Yeah, I don't know why they wouldn't time correct it or offer a timing correction setting. Seems shortsighted and non optimal.

[@samsonstrike]

THX!

[@DaveRat]

👍🤙👍