The daw war is over. midi timing! woooooo!

Hex,

You've stated a bunch of claims which, IMO, warrant further exploration before they can be accepted as fact or opinion. MIDI timing is a subject of interest to me, so I'm responding here as I had in your previous thread ( http://discussions.apple.com/message.jspa?messageID=5353527#5353527 ) where it seemed you were encouraging a dialog on the subject, but you haven't replied yet.

In your previous thread I offered a critique of your test method and I think it would be interesting to discuss it in more detail.

Aside from that, I'm not sure that MIDI timing was so terribly full of "jitter" in Logic 7 anyway. And it stands to reason that different MIDI interfaces would exhibit different degrees of accuracy. So if you're up for a discussion (this being a discussion forum and all), I'm down...

Logic HAS BEEN the Midi king ever since Notator:-)))))

For programming midi, Logic wins hands down, for ease of use, again logic wins....

But for playing back exactly what you program in with no differrence in timing or quantize issues, I'm afraid Logic Loses. I don't know what it is, but I had ischwartz helping me out through these boards trying to get to the bottom of it, but Logic just doesn't play back exactly what you play in via midi. It adds a weird stiffness or quantize to it even if there is none added....
I am about to use ishoot so that I can put a video on youtube, as I need people to see what I mean...

Try prrogramming a real lazy beat, almost behind the beat, like Dre style, dont quantize it but listen to it back as soon as you recorded it....It's not the same thing!

Digital Performer on the other hand, does play back exactly what you put in.....Don't take my word for it, but watch the vid when I post it!

+*ut for playing back exactly what you program in with no differrence in timing or quantize issues, I'm afraid Logic Loses. I don't know what it is, but I had ischwartz helping me out through these boards trying to get to the bottom of it, but Logic just doesn't play back exactly what you play in via midi. It adds a weird stiffness or quantize to it even if there is none added....*+

That is because the midi gets quantized even though quantize is set to off... Midi in Logic(and other DAWs) is tick based and the maximum resolution you'll get in Logic is 3840 ticks within one bar... Logic runs a steady and NONE Tempo based time...(Unless you program time tempo changes) that means that if you play a piano line Logic has 3840 grids that it can quantize your notes on.. This is not enough... As an example the MPC 4000 has either 10000 or 40000 ticks within one bar (If I remember correctly 10000 ticks is the quarter bar max reso with no quantize)

I've been doing a lot of experimentation with this. If you wanna conceive 100% of what the artist was playing you'll need to record straight to audio or use an MPC4000. Even if you quantize the MPC it will still sound VERY natural as the secret to an MPC is the playback algorithm. I have analyzed the MPC during playback to find out why it's timing is the best available... Its very simple:

Lets say the MPC has a tempo of 100. When you hit play it will start to play but analyzing the tempo will look like this within a second: 100.002, 100.005, 99.995, 100.004 etc etc.... So the MPC is varying the tempo in a micro way - just like a human being would do it as NO human being can keep a tempo 100%... Logic can keep a BPM of 100 to 100%... That is why a person with a sense of timing can feel that something is NOT right... or static...

For some purposes - I'm sure - static is wanted. I just personally dont like it. So I switched to program and record ALL midi in my MPC4000. It's pretty easy if you dont wanna do a lot of editing and if you are a good player and the MPC can still utilize all the great VI'S in Logic.... This way I am also taking a load off of Logic...

My Setup is this:

Logic is my Sample Player and hosts ALL virtual instr. but is doing NO playing...
The MPC is handling ALL midi and arrangement
Protools is recording/Mixing and monitoring everything...

All done on the same G5

Steve

well you actually do... Q is whether you hear it or not.(not meant as disrespect - it depends on your musical style)

The thing is that within one bar Logic is quantizing though quantize is set to off - and it quantizes to 3840 tick per bar.... That is a pretty rough quantization - if you wanna keep a live feel... It has NOTHING to do with your midi interface....

You can verify this yourself:

Select a midi region and now in the inspector set quantize to off - and it will show you off (3840) - you can also select a note in the matrix editor and then open the event list. There you'll see that the max you can nudge a note is 240 ticks per 16th bar (240 x 16 = 3840)

Yup - Logic is THE midi DAW...

if you want robotic static things another cool way is also Logic's step SEW (The Plug) or FxPansion's GURU...

SteveDjokes

Your a genius!!!!! Thanks mate, just the information I wanted.....I knew I wasnt going mad!!!!!

It's a shame that DAW's cant do this....

Many thanks.

You're welcome 🙂

guys, of course there's a solution which we used to use with old hardware sequencers back in the day.

double the tempo and programme 2 bars for every one bar = double the max' quantise resolution then with logic that'll give you 7680 ticks per 'bar'

make sense?

once you've made the 'lazy' beat you can render it (or render individual drums is even better) and reset Logic back to 1-bar = 1-normal-bar and reset the tempo

anyways logic has always had the best timing for midi imo on PC, thats why i started using it in the first place cos vst 3.x was still too shakey for dance timing back then and Logic offered a stable clock... never looked back frankly. Still got 2 1040's with cubase tho just for nostalgia 🙂

Let's put this in perspective. You got 960 ticks per quarter note, so the size of a tick depends on the tempo. At 120 bpm, a tick is just over half a milisecond in length. Faster tempos it becomes shorter. Only as you get down to around 62 bpm do you have a tick - and corresponding "quantization" error - roughly one milisecond long. So it's true, even when quantize is off, logic is moving your stuff around by (usually) less than one milisecond. In the context of this thread about midi jitter, we don't even have hardware capable of resolving to this accuracy or measuring it. 960 ticks works more or less the way 44.1 hz works or 24 frames works - It generally exceeds our ability to percieve.

Try putting a sound with a good transient, maybe a woodblock, in your delay, equal dry/wet, no feedback. Start at zero and gradually increase the delay time. How many miliseconds before the effect stops sounding like phasing and begins to sound like the slightest flam?

We can't rely on midi to solve problems of phase coherency, stero imaging, etc. Even speaker wire and amplifier circuitry are barely up to that task. MIDI harware is nowhere near capabilities under a milisecond. Whatever the timing problem, 960 ppq is not it.

Hexfix,

I'm using a Unitor 8 and an AMT8. But the notion that Logic 7 has timing "jitter" that's in the 3 ms range doesn't jibe at all with my experience. I'm still not convinced that you arrived at an accurate test procedure to assess this.

Rob, I don't remember what situation I was helping you out with, but if you can recall the thread I'd be happy to review my thoughts on Logic's timing at the time.

Steve, you are right that MIDI data recorded "unquantized" is actually quantized anyway (to the tick resolution of the DAW). That's the way it's always been. So the degree of accuracy of MIDI recording/playback of live-played parts is dependent on TEMPO. If you want more accuracy in your MIDI timing of live-played parts, double the tempo before you start recording. However, the accuracy of playback of QUANTIZED parts will not vary with tempo. I've got a 20+ year history in dance music and I've never had a producer complain about flamming or phasing or lack of lock because of some "jitter" in MIDI timing during the playback of drums and loops out of Logic. Never happened.

There is only one way to accurately test the MIDI timing of a DAW, and that is to record the MIDI signal itself as audio. This can be done without test equipment, supposedly, by taking a MIDI cable and putting a 1/4" end on it. Record the electrical "blips" and count samples between stop bits.

In any event, the only way anyone can really know how accurate the MIDI output of a DAW is is to analyze the timing of the MIDI data itself --- not how a target device reacts to it. See, even if a DAW has good MIDI timing on playback, there's no guarantee that any external MIDI device will play that data back in time correctly. That's because MIDI is an asynchronous system. So when a device receives a MIDI message, when it's actually decoded and then plays a sound with respect to the timing of the DAW's clock all depends on what point that device's microprocessor is at in its duty cycle --- not to mention how much other MIDI data might be present in the stream that the device's microprocessor has to otherwise act on or ignore.

Running status plays a big part in how well external MIDI devices react to MIDI in general too...

It's a near miracle that any of this stuff works at all to produce good playback, yet it does much of the time! If there were severe jitter (and 3 ms is a lot of jitter), Logic wouldn't be usable for playback of anything.

And finally, Steve, I don't buy that you've been "testing" anything. Just like you said in that other thread you tested the latest time stretch algorithms and that they sounded better to you (when in fact they haven't been changed at all?) And if you don't know whether an MPC is recording at 1000 or 40000 ticks? Pretty wide gap... makes me wonder how you can test something if you don't know the quantities involved. And how are you judging the resultant MIDI timing?

Steve, you're knowledgable in many respects of Logic, DAWs, PT, etc. But if you're going to discuss this subject you should at least have your facts straight.

Message was edited by: iSchwartz

Charlie the Tuna is absolutely right, as is iSchwartz. The MPC-4000 actually has the exact same MIDI resolution as Logic, that was why I ultimately bought Logic over the 4000. When i was using the MPC-2000, it only had 96 ppq, and I could definitely hear small quantizations.

Both Logic and the 4000 have 960ppq, unfortunately I haven't been able to test and see if I can hear if Logic is quantizing my MIDI data due to interface issues.

You should NOT hear any quantization with 960 ppq. If users are having these problems, it really should not be the 960 ppq MIDI resolution of Logic.


By the way, a good rough way to test this is to see how a sequencer handles recording of snare rolls from a physical electronic drum kit.