New Audio Engine Request Logic Pro X

I'm happy for you, but still your claim about midi timing is not correct. If you can share with us a project exhibiting that problem I would be happy to take a look. you really have no idea about me any more than I have about you. You have made a series of insults towards me in the past hour for what reason I have no idea, I guess because you don't like being challenged but you have no basis to state what I know or don't know. I humored you to make a test on my system and found no problem, including even doing a null test to verify it. Your claims are completely unfounded. Time for me to go to bed now... have a nice evening.

I love to be challenged. I wouldn’t be in this industry if I didn’t like one. I have watched Grammy award winning mixers post questions like this online, and someone with your exact response ALWAYS responds. They don’t even leave enough time to pretend their testing it, and they come back saying “null.” Pretty hilarious when someone actually has time and posts their observation. Ironically the nuller always back tracks and blames his own pilot error. I love logic and use it, have no need to make up anything. I would only be doing you a favor by showing you, but I really think if it’s one of those things that you can’t notice, then you’re good. Some people are just a little more meticulous.

If it makes you happy. I’m wrong and youre right, go Make music.

I've talked to people that I respect professionally who concur with Crysler's observation on MIDI timing. Some have gone to great lengths in testing this, using external hardware, and then either keep using Logic for convenience or use another DAW.

Personally, I find Logic's timing unpredictable. If a DAW's metronome and timing can EVER audibly slow down at times, then it's not hardware based rock solid, no need for further evidence. And I do get these drifts, on and off while working, I don't care what others say. I'm sometimes startled by a kind of rubbery feel. Sometimes changing clock sources helps. But then it's not likely that Logic can always deliver perfect timing.

I'm a C/C++/Assembly coder and hardware designer. Whether this is in Logic or CoreAudio or macOS or Apple hardware is moot, for me. Irregular response to hardware events is why macOS, its admirable reliability notwithstanding, is not a real-time OS that can run process control or life support systems. Granted, if such a desktop OS can support a non-maskable interrupt queue for drivers to process a few high priority events, it can nonetheless come close, but it takes skill. The recent T2 fiasco shows Apple lacks such skill, or considers it non-critical, or even skips in-depth testing of timing.

It happens that I like the Logic app enough to use it anyway and make the best of it.

@Dewdman42, if you care, can you try the test Crysler is suggesting?

I don't think anybody is lying here. The OP is clearly not technically proficient. But SOMETHING more than a placebo effect is making some systems have more timing slop than others.

If it's more than a placebo, surely someone can show us an example?

IMHO it's unrealistic to expect a bloated OS, that's meant to be connected to the Internet to serve up your preferences and habits to advertisers and other interested parties, to deliver sub-millisecond timing in your audio. That's a bit like performing brain surgery in a circus.

My studio machines are off the 'net > 99% of the time, but Siri won't even tell me what time it is, and a number of important utilities randomly beachball trying to call home. Phones are for that, not computers.

@David, let's agree that it's usually reported as inconsistent and intermittent. That doesn't make it imaginary.

One more detail is that if the timing drifts relative to say an atomic clock, it can still appear perfect to the OS and all the the apps that are drifting and jittering together. The effect on output audio and mixes is hard to predict.

I've thought of catching drifting timing on video, but by the time I find my phone it's gone. It doesn't matter, if you aren't affected, you're in good shape.

Actually, not only can it easily be "imaginary", it turns out that our ear and brain are designed from the ground up to do that on purpose....for survival reasons. Here is a great video from AES that explains it...

much ado about nothing has been made many times trying to claim many things, including this DAW is more transparent then that DAW...and the truth is....legitimate tests with equipment that don't imagine things, have verified many times that all DAW's are equally transparent in their raw basic audio engine, not withstanding that people have imagined it otherwise due to the way our ears work and a lot of marketing hype. What you do with the DAW is what matters.

Regarding timing, if anyone can provide an example, then it would be good to get to the bottom of it. External MIDI I would not expect to ever have anything even close to sample accurate timing, especially with the advent of USB interfaces. If the interface is MTS compliant, then it should still be sub millisecond and good enough to sound tight, but will not be sample accurate. Software instruments will be sample accurate every time.

All the MIDI timing tests have been extensively done ( check Gearslutz ). Under good conditions, Logic was shown to have comparable (ie basically the same) timing and jitter as the gold standard for people preaching rock solid MIDI timing, the Atari STs hardware MIDI ports.

Now, that’s with good hardware and drivers that properly support OSXs MIDI time stamping. Some MIDI interfaces have poor drivers, or don’t support time stamping, and the MIDI performance here is compromised (note that this has nothing to do with Logic’s timing, but the drivers).

There is one case where Logic itself wasn’t timestamping MIDI events, and I haven’t checked the see whether this is still the case, I suspect it is, as it seems to be the nature of the Environment - so make sure you use the External Instrument plugin to do MIDI timing tests.

What was the next problem? Oh, how somehow Logics summing engine is adding high end rolloff and low end smearing when it adds those floating point numbers together, to sound dark and muffled and wrong in the low end?

Again, numerous properly done null tests have shown there is basically no difference in summing between DAWs. The differences only start coming in when you add in differing factors, ie pan laws and other processing that varies between DAWs, and these will contribute to your mix results.

Now, by all means, if you aren’t getting the results you want from DAW 1 and *do* get the results you want on DAW 2 then fine, use whatever gets you there. But whatever DAW you pick, *someone* will be saying it doesn’t sound good as *insert name of other DAW*. Without posting examples that demonstrate your claims, on this well-trodden ground, I tend to assume these days people are talking emotionally, rather than factually...

There is one case where Logic itself wasn’t timestamping MIDI events, and I haven’t checked the see whether this is still the case, I suspect it is, as it seems to be the nature of the Environment - so make sure you use the External Instrument plugin to do MIDI timing tests.

.

If you happen to remember or find out what that case is where timestamps aren't getting made, I'd really like to know also.

I have also been suspicious that LogicPro is not reading timestamps on incoming midi, since it goes through the environment before hitting the sequencer. But I have no evidence or data to back that up. Don't even know how I could test for it.

Basically, the environment is basically operating real-time on MIDI events passing through, and just like standard MIDI, there is no concept of timestamping. So if your track is outputting through the environment (ie through a MIDI environment object) you get no time stamped events. but if you use a software instrument track with the External Instrument plugin to output to your MIDI hardware, you get time stamped events.

It’s very easy to see these with a MIDI monitor like MIDIpipe or similar, which will display the Core MIDI time stamps if they are there.

Like I say, I haven’t rechecked recently to see whether this behaviour has changed...

Before everybody goes home with their usual beliefs, I'd just like to make sure we distinguish between two very different things. Let's not mix them up.

One is claiming that Logic has a specific "sound". That has been debunked, at least for simple configurations where no effects are used, by precisely comparing output files between different DAWS using the phase inverted "null test".

The other is questioning the rock-steadiness of Logic's timing, under load, relative to an external system. Different issue. Though jitter and drift might at some point impact the "sound".

right I understand that. Same thing goes for midi coming from an MTS-compliant interface though...it has to go through the environment before it hits the sequencer object. So are timestamps lost on input into logic?

I've not done any testing in MIDI input, so I don't know the exact behaviour.

As far as I know, Logic simply records the event at the position the sequencer was in when the event was received. Thinking about it - I'm not even sure how timestamping would work on input. The whole idea about timestamping is that when the sequencer has a bunch of events to fire out, it knows where in "time" (referenced to it's playback) those events should occur, and can time stamp the events and send them to the interface ahead of time (assuming the interface supports time stamping of course) and trust that the interface will pass them on as accurately as possible.

On MIDI *input*, the sequencer knows nothing about the "intended time" of the incoming events - how would the interface time stamp an incoming event? The interface has no concept of where a particular sequencer application might be in time, it simply receives events and passes them to the system, where they get picked up by whatever apps are listening to that MIDI port.

I've always assumed that MIDI time stamping was more of a "playback" type thing, designed to improve the accuracy and congestion of output events (this is what the old emagic AMT system was, for example, it simply made sure that events that were due to be played back at the exact same time from different ports actually left the ports simultaneously, rather than serially, for better timing. OSX's time stamping is not the same tech, it's designed by the former Opcode guys).

So I'm not sure time stamping is a thing on MIDI input from something like a keyboard via an interface at all, but someone else might know better...

Mts is based on a very low level nanosecond clock, with a running number that is always increasing over time. Mts is not awareness about beats and stuff like that, it’s simply a micro clock that is running always. When you see the mts timestamp in an external logger it will display as some kind of microsecond number that just constantly grows over time. Then the midi interface needs to synchronize itself with that running clock in order to know when to play timestamped events so that it’s own internal clock will determine when precisely each event will go. The main point of this is not to schedule events huge amounts of time into the future it is simply to eliminate jitter caused by an extremely non-real-time operating system and usb driver.

Likewise when you play your keyboard you want to elimate that jitter. The clock sync can work in both directions, why not? If you monitor events from your keyboard in another product that will show mts then you can find out if your midi interface is stamping mts or not and coremidi is supposed to pick that up.

The question is whether logicpro uses it is not testable in any way that I can figure out.

Regarding history, way back when there were two interfaces that supported mts, each their own way. Emagic and Steinberg. I don’t think opcide ever produced an mts interface but my aging memory could be off. At some point Apple built mts of some kind into coremidi. Then it would be up to the makes of each interface to code their drivers to use coremidi mts in its driver. A particular interface may or may it be using it, and it’s all handled in the driver as far as adding timestamps to midi events as they come in. I suppose core midi or the driver might stamp them with a number even if the interface isn’t providing it, but then it would have inherent jitter due to the driver’s non real-time nature. Interfaces today do not tell you whether they are using mts any more. I asked motu once about theirs and their support finally told me that their interfaces are compliant with coremidi mts, whatever that means, we never hear the details and it’s really hard to test for it, especially on the input side.

But jitter matters in both directions.

When you see the mts timestamp in an external logger it will display as some kind of microsecond number that just constantly grows over time.

Yep, as I mentioned, it's easy to see with appropriate tools. I've also done a certain amount of Core MIDI programming as well...

Then the midi interface needs to synchronize itself with that running clock in order to know when to play timestamped events so that it’s own internal clock will determine when precisely each event will go. The main point of this is not to schedule events huge amounts of time into the future it is simply to eliminate jitter caused by an extremely non-real-time operating system and usb driver.

 

Likewise when you play your keyboard you want to elimate that jitter. The clock sync can work in both directions, why not?

Makes sense, yep.

If you monitor events from your keyboard in another product that will show mts then you can find out if your midi interface is stamping mts or not and coremidi is supposed to pick that up.

Ok, so I just used MIDImonitor with what I had to hand and the incoming MIDI events are indeed time stamped, using an NI controller keyboard via USB.

I'll have a look at some other interfaces later (eg, emagic AMT8 via USB, Focusrite audio interface etc...) to see whether they time stamp the data too.

The question is whether logicpro uses it is not testable in any way that I can figure out.

Indeed, and I'm not sure how it would be possible to easily test this?

Well, you can get an idea of the kinds of jitter figures you get with properly supported time stamping, versus more jittery data without time stamping, and then timing tests would give you an idea of whether the jitter performance suggests any given driver is properly handling time stamping or not.

The jitter tests across a range of hardware with Logic over on the GS forum are quite interesting, if you haven't read that...

See the posts from Jauqq in the thread:

https://www.gearslutz.com/board/electronic-music-instruments-and-electronic-music-production/957824-free-software-measures-midi-latency-jitter-5.html

Yes, this was all jitter and time stamping performance with MIDI output - like I say, when I was looking at this with him, we were investigating only MIDI out performance, which is how we discovered that environment MIDI out had no timestamping, but external instrument MIDI out *did*, and performed much better.

Edit: I just checked, and Logic still doesn't output time stamps on anything from the environment, so for the best MIDI timing, use the External Instrument for external MIDI gear, not regular environment objects.

That's only because you packed 128GB of RAM

into that cheese grater of yours, Dewd