MIDI timing discussion

t · 2017-05-14T16:51:14.046Z

DT/MPC vs. DAW and the rest of the world

So my opinion is based on generalities…

You’re entirely entitled to your opinion, but once you actually start testing stuff, you rather quickly find out that:

The MIDI clock output by Elektron gear is rather jittery.
The MIDI clock output by Ableton Live is a lot less jittery that you might assume.

We all just assume that dedicated hardware is better in terms of clock stability, but that really isn’t always the case. I was pretty surprised when I found these results, and actually went through the trouble of getting a second test device just to confirm my findings (it‘s all in the thread at http://mutable-instruments.net/forum/discussion/comment/129861#Comment_129861 for you to read and verify).

LyingDalai · 2017-05-14T17:07:49.073Z

Indeed, I should have done my homework
So my intuition might be wrong.

There’s also this website that reports all kind of measures but it’s hardly as readable as your link…
http://www.innerclocksystems.com/new%20ics%20litmus.html

t · 2017-05-14T17:28:25.825Z

DT/MPC vs. DAW and the rest of the world

So my intuition might be wrong.

In all honesty; I was as surprised as your are.

I always assumed that hardware gear would be much more stable and that software would be much more jittery. You can still see my surprise on the first page of that thread when I first test my MPC 1000 back in April 2015 just after finishing building the testing device…

DT/MPC vs. DAW and the rest of the world

There’s also this website that reports all kind of measures but it’s hardly as readable as your link…

Yes, I am aware of that page. I’ve just been testing MIDI Clock output while these people also seem to test the internal clock stability. However, they don’t publish their procedure or any way to verify their results independently so I’ve always taken their results with a big grain of salt.

Open_Mike · 2017-05-14T17:31:50.627Z

No offense anyone, feel free to test away, this is a music geek forum after all. I however just use my ears and am experiencing very tight sync between Elektrons. A few posts from the linked pages start with “I thought my gear was tight until I tested it”… I imagine most of these machines get tighter sync than actual humans and then we adjust them with micro timing and swing to feel looser again… I also am not completely dismissing the sync gen sites reports as well, but really even those don’t matter to me. If the sync is not tight my ears tell me, If my ears don’t tell me I’m quite satisfied and can care less about millisecond on grid prescision… YMMV…

Open_Mike · 2017-05-14T17:39:58.973Z

The trouble is that a lot of Ableton users “hear” thier gear not synced up tightly. Could be all sorts of reasons with buffers, delay compensation, midi interface, audio interface, drivers, etc…
When hooking hardware together and monitoring from an analog mixer, often the gear “sounds” tight…
All testing aside…
We always hear of daw/ableton users looking for solutions such as sync gen, usamo, midigal, etc, because their ears tell them something is off…
Rarely if ever do we hear about someone looking to put something in between their OT and other Elektron because it doesn’t sound tight enough to them…

pidgeo · 2017-05-14T17:59:40.857Z

yea i always have to do some serious tweaking to track delays and such with ableton hardware sync but then it’s usually pretty solid.

t · 2017-05-14T18:48:03.158Z

DT/MPC vs. DAW and the rest of the world

I however just use my ears and am experiencing very tight sync between Elektrons.

That‘s correct. There are two ways in which a device can sync to external MIDI clock:

When using external clock, the internal clock is not used and the sequencer progresses with every clock message received. In this case a resolution higher than 24 ppqn is not possible, and any swing or “micro-timing” would depend on the clock signal. This is what older and “simpler” devices do.
When using external clock, the device derives a tempo value from the received clock messages and uses that to sets the tempo for its internal clock. This is what most modern or more complex devices do.

You Elektron boxes use the second method. Because in that case the tempo is derived from averaging multiple clock messages, any jitter in the clock messages has no impact on the stability of the internal clock.

t · 2017-05-14T18:52:46.483Z

DT/MPC vs. DAW and the rest of the world

The trouble is that a lot of Ableton users “hear” thier gear not synced up tightly. Could be all sorts of reasons with buffers, delay compensation, midi interface, audio interface, drivers, etc…
When hooking hardware together and monitoring from an analog mixer, often the gear “sounds” tight…
All testing aside…
We always hear of daw/ableton users looking for solutions such as sync gen, usamo, midigal, etc, because their ears tell them something is off…
Rarely if ever do we hear about someone looking to put something in between their OT and other Elektron because it doesn’t sound tight enough to them…

Exactly. The problem most people experience has to do with inevitable latency issues when trying to synch up ITB sound sources with external devices through a mixer, or even worse, routing external signals back in.

It’s easy to blame these issues on the MIDI clock and some vendors have indeed tried to monetise on this. I really don’t like that because in most cases the problem can be fixed by a better understanding of the latency issues inherent with computer audio, and by getting your settings right.

PS. You mentioned the MidiGal project in your list. Please note that it’s just an Open Source / Open Hardware platform that allows you to run various utilities. In in no way claims to fix clock problems.

t · 2017-05-14T19:43:08.167Z

DT/MPC vs. DAW and the rest of the world

“MidiClk is a high precision MIDI Clock tester/generator which allows you to check MIDI Clock quality with great precision and generate super stable MIDI Clock synchronous to the incoming MIDI Clock. With MidiClk you can finally use your DAW as a clock source for hardware sequencers, arpeggiators, and synthesizers!”

Ah, yes. It can indeed behave as I described above under point 2 and then use the derived tempo to generate stable MIDI clock for a device that uses the method I described under point 1. I never used it for that, but in that sense it might indeed be able to fix clock issues.

LyingDalai · 2017-05-14T21:43:04.755Z

Only moments where I let Ableton give the tempo instead of OT is when I jam once a week with my friend Tamaris.
He plays on Ableton Suite + APC40, I record every track on my own Ableton, and being able to sync each other via Link (on local WiFi) and have both the hand on the tempo is such a pleasure !
In a few years all these MIDI and USB cables will have disappeared, and it will be such a pleasure !!

umonox · 2017-05-14T22:15:17.288Z

WiFi? what a nightmare! may be fine for those with money who life in their own house - but for the majority of people (like me) living in huge (or even smaller) 35 to 60 m² apartment blocks, this will never be an option. unstable like hell due to literally thousands of different radio waves from routers, smartphones, tablets … influencing each other … no - i prefer cables! cables ftw

LyingDalai · 2017-05-15T05:12:34.998Z

Works pretty great at my place, I live in an apartment but it’s not as crowded as Paris
Nothing to do with money, all you have to do is live in smaller/cheaper town
I guess we will also see Bluetooth sync…
Once you tasted wireless sync, hard to go back.

umonox · 2017-05-15T06:22:08.840Z

i live in a smaller town too. prices for ground raised significantly in the last couple of years. building businesses charge insane prices for building a house too - because in Germany craftsman is the last job the current generations would want to learn. they all rather study some usless sh*t because they are in the wrong belief that studying automatically leads to much money yeah, but thats why those companies have constant skilled worker shortages and therefore charge much more per hour. taxes and requirements for building standards are also high. so - a house is something insanely expensive in this country here. unless you inherit it

yeah and in our apartment block we have lots of families and the kids all have a smartphone, the parents have smartphone/tablet, gaming consoles all connected wireless. also lot’s of singles here which have calls all day and night - and lots of people without a job too - they all use their stuff all the time. and the apartment block is built with old standards, pretty old house actually. so no electromagnetic shielding from the materials used to build that thing. if i fire up my old hifi system (which isnt shielded) i hear the sound of those radio waves - pretty much all the time and whereever i put this thing my wifi connection is off. i tried it once, but every here and then it disconnects me. no surprise.

oh and one other thing: i heard from DJs getting hacked in the middle of a set - they were using wireless sync too i would never want to risk something - if i would be a DJ of course

void · 2017-05-17T09:22:35.000Z

DT/MPC vs. DAW and the rest of the world

Get two computers with Ableton. Do not connect them with a synchronization cable. Press play on both computers at exactly the same time. Leave them to run for a year. (switch off software update etc etc.) When you return after a year they will still be running within a ms of each other.

nah man, definitely not.
personal computers are actually pretty bad at keeping time. pushing out audio at 48kHz on one computer will be slightly slower/faster than on the computer right next to it.

also, you said it yourself - switch off software update etc - there’s ALWAYS the possibility of your OS just doing any random task you have no control over and which might hang/delay something. The OSes we’re using aren’t realtime OSes.

MichalHo · 2017-05-17T10:03:13.411Z

This is not true.
I have never encountered a stable computer clock (osx or windows) in 25 years of working with computers and music.
The only reasonably stable master clocks in live situations (live sets up to 7h long) are older MPCs or Octatrack. This comes from years of first-hand experience and is also easily explained: the system architecture is streamlined for this job, unlike personal computers, which will always have intrusive processes running somewhere, no matter how much you tweak your settings.
The more laptops in a live/stage setup, the more resyncing you will have to do during the course of the night.
External clocks won’t change that, the weak link is the OS.

MichalHo · 2017-05-17T10:12:39.392Z

Sorry, but Ableton regularly proves you wrong. Try measuring over a period of many hours.
Also… you generalize ‘elektron gear’ but mean AK and A4. There is no mention of OT in this ‘test’.

MichalHo · 2017-05-17T11:11:26.846Z

No offense, but you are going off on a theoretical tangent that is barely on topic.
I’m interested in the stage/performance context and MIDI clock, not Caesium clocks, not USB.

MIDI is serial too as you know, but it generally avoids the packet issues of USB. So is that protocol getting the atomic clock data to your clock instance. Let’s forget about that, as that discussion does not actually solve any sync problems.

I agree with your last statement, which, had you made it in the post i previously replied to, would have made my reply redundant.

t · 2017-05-17T11:49:28.641Z

DT/MPC vs. DAW and the rest of the world

Sorry, but Ableton regularly proves you wrong. Try measuring over a period of many hours.

Actually did run the initial tests I did overnight as well because I was so surprised with the results. With Ableton Live on my MacBook Pro, there was no difference between running the test for a minute or for 12 hours.

DT/MPC vs. DAW and the rest of the world

Also… you generalize ‘elektron gear’ but mean AK and A4. There is no mention of OT in this ‘test’.

It’s indeed possible that the OT has rock solid clock output, but looking at the results from other hardware gear I and others tested with this, I strongly doubt it.

Please note that both the hardware and the testing firmware is entirely open source. You can easily build one yourself or have it done by someone with the required soldering skills for about $30 - $40 in parts. See MidiGAL | MidiSizer for details.

You can find more information on the testing firmware here MidiClk clock tester/generator | MidiSizer

This means that you can easily replicate the results yourself if you like.

MichalHo · 2017-05-17T16:28:37.855Z

Thanks for the concise response.

Lab condition testing is fine as a guideline, but take a typical live situation:

You are running ableton, jamming in session mode, you are sending midi clock to one or more musicians that use ableton-laptops or hardware. Sooner or later, while using multiple instances of a CPU intensive plugin like e.g. turnado, your computer will have a ‘hiccup’ and miss a beat. Ableton’s midi clock will slow down and speed up again briefly, which will not be neutralized (or resynced) by the receiving devices. Within a split second, your co-musicians might be off by a quarter note, you might get a ‘snares on the one and three’ scenario. Everyone will have to resync to the first beat. I see this kind of thing happening on a regular basis. Ever since I’ve started insisting on being master clock with the OT, this problem has gone away, downstream to the slaved devices, to be precise. Hardware usually stays synced, laptops still have their hiccups, but not all clock recipients are affected by a master glitch…

Tbc. Off to dinner.

Ps. You could replace the term ‘hiccups’ with drop outs. The clock ticks might simply stop for a critical moment instead of changing speed too quickly.
In this case the mechanism of sync dysfunction is not as important as its result.

orangehexagon · 2017-05-17T17:53:46.742Z

i think we have to be careful about stating intuition or opinion as fact. the spread of misinformation these days is rampant.