I noticed some time ago that when you record external MIDI instruments into Logic that the resultant audio is delayed. This was expected (although it is a positive delay which is weird) and should be easy to overcome by adding delay to all MIDI signals or the track delay on the arrange page or a recording delay in the core audio preferences, by the same amount thereby bringing the audio back into line; similar to any other sequencer. However in Logic the delay is not constant so this is impossible to do.

I've posted on here and several other forums but never had a response. Today I found this thread on Logic Pro Help Forums which confirms my findings (and my fears).

So I have a question for all you using external MIDI instruments with Logic... HOW???.

Sure if the audio starts with sharp transient you could pull them back into line each time, although that's not ideal anyway, but what are you all doing with non-transient stems? Are ALL Logic users 'in the box'???

yes, I have had this problem.

It only seems to be a problem when I have lots of plugins loaded. especially UAD plugs. It is not a problem when I don't have many plugins loaded, or are only using logic's own plugs.

I never really had time to investigate the problem fully, but I guess it is a bug with the Plugin Delay Compensation. If I get some spare time I will do some testing.

--------------------
www.Lozjackson.com

Use low buffer sizes for recording, and/or jiggle stuff around to get the feel you require. Also understand how PDC works and what it does for other timing dependent things. Run some tests if you're not sure.

Sometimes it's very easy to get caught up in the "OMG! My recorded guitar part is 15 ticks late!" when the timing inaccuracy of the notes you play might be *way* bigger than that.

You just need to know how this stuff works and especially the quirks that often aren't really manual topics, and develop a workflow that works for you.

And another reason some of this stuf gets overlooked is that where once MIDI instruments were a required necessaity for making electronic music, nowadays more and more people are junking their MIDI hardware, and a whole new generation of computer-based kids are working without ever having touched another instrument, let alone understanding what MIDI is.

Yes, in an ideal world you shouldn't have to do all this, and everything should be perfect - but the reality is life is one long series of workarounds...

(Or at least it feels like it sometimes...)

Bossman, it's a problem for me (and everyone over at Logic Pro Forums) everytime regardless of plugins loaded so I think you must be one of the lucky ones. Check those screenshots in the link I posted and feel free to run the test yourself, I guarantee you'll get the same results. And it's nothing to do with PDC as I have it switched off.

Desmond I really appreciate you taking time to respond but I dont think you have understood the situation. If you know a 'series of workarounds' for this problem I would be very grateful to hear it (which was exactly my question in the OP) otherwise, thanks anyway.

I understand the situation perfectly, thanks.

By workarounds, I mean ways of working around any undesirable behaviour to find the best way forward, not necessarily making the software behave differently to how it currently does.

In this case, it means recording using low buffer sizes, not recording when you have latency inducing plugins delaying the audio paths, checking your timing on critical parts (ie no one's going to care about a 35 tick delay on a pad part with a slow attack, for instance) all all the other usual things (calibrating your record delay offsets and so forth), and generally understanding and paying attention to timing once you know what's going on.

And of course, sending comprehensive test and bug reports to Apple feedback in the hope that a magic Logic update will fix/improve this behaviour.

But thinking that no-one could possibly use Logic to record music because of this is a little extreme...

Quote desmond:

---

...thinking that no-one could possibly use Logic to record music because of this is a little extreme...

---

I appreciate people are using Logic to record external MIDI instruments, which is why I am asking the question... HOW?.

You dont give me any confidence you know the answer, no offence.

If the answer is to reduce buffers to 32 each time you want to record, as suggested in that thread on Logic Pro Forums, then fine, but that dont sound right to me, is that really what everyone does? You have to reboot Logic everytime you change buffer settings!

Quote stinkfinger:

---

I appreciate people are using Logic to record external MIDI instruments, which is why I am asking the question... HOW?.

You dont give me any confidence you know the answer, no offence.

---

I've been recording MIDI instruments using Logic for 16 years. Sorry if that gives you no confidence - not a lot I can do for that. All I can do is give you my perspective and opinion on the issue you raised, and tell you what I do, which is was you asked for in your original post.

Note that this timing jitter, at low buffers sizes anyway, is in many cases a lot better than the MIDI response times of many MIDI instruments anyway. Some/many devices are very sluggish between recieving a MIDI event and generating a note - so don't think that even if Logic was perfect and didn't exhibit this behaviour, that all recorded notes are going to be hitting perfect 16th note quantised transients anyway.

MIDI has very poor timing in general - Unlike the sample accuracy of plugins, the best you can do with MIDI is send a note event at basically the time you want to hear it, and hope that the audio from the device happens relatively quickly. With small data streams, it's mostly good enough - with larger ones, not so much.

And people have been making MIDI music since the 80's, so...

Quote stinkfinger:

---

If the answer is to reduce buffers to 32 each time you want to record, as suggested in that thread on Logic Pro Forums, then fine, but that dont sound right to me, is that really what everyone does? You have to reboot Logic everytime you change buffer settings!

---

No, you do not have to close and reboot Logic to change buffer sizes, Logic simply needs to reinitialize Core Audio.

But yes, this is not something you do not want to be doing constantly. When you are recording, use low buffer sizes. When you are done recording and have moved onto editing and mixing, you can up the buffer sizes to something less stressful on your machine if you need to.

If you workflow means that you don't work like this, you are constantly recording/editing/mixing until the end, then it makes sense to stick with low buffers all the way through the process. And if these timing issues are really bugging you, then when you print your MIDI instruments, do a quick timing check and nudge if necessary.

I can't speak for everybody, and I don't know the "answer" of magically making Logic do something differently to what it currently does (I can't get the devs to magic up a personal update just for me), so if you are waiting for someone to show up who can, then you might have a long wait.

Not much more I can add to what I've already said... *shrugs*

I totally agree with Desmond .
I've been making music for over 20 years now , and it's always been struggle to get machines to do what you want them to do .
Coming up with workarounds for the limitations of the gear is simply part of the workflow .
Analog synths used to get out of tune all the time , tape recorders being synced with smtp got out of sync , the one machine was great at this but couldn't do that , etc .....

How comes people nowadays expect everything to work perfect ? It never did and it never will . When you try to connect totally different types of instruments from different manufacturers and even different generations , there's always going to be issues .These days you get a great set of tools for very little money , so just try to be creative with it ... it's part of the fun ... :-) ...

--------------------
mac pro 8 core 2,33GHz , macbook pro 17" 2,4GHz , logic pro 9 , 828mkII , NI complete 5 , all spectrasonics , mackie control universal , genelec .........

To the OP, you need to familiarise yourself with the "Event Float" window inside Logic. Its functions allow you to move regions, i.e your recorded midi events, in both directions, at a resolution of 960 ticks per quarter note.

Correcting timing inaccuracies by ear, using the Event Float, is quite easy and straight forward. It's also fast.

@Spacerabbit, thanks for your little 'kids these days' speech but ultimately nothing of any help whatsoever in there, what was the point of your post...(rhetorical).

Tui, I'm trying to record stems from my MPC, I'm fully proficient with the event float and am aware I can pull the stems back into line manually by ear but it would be nigh on impossible to retain the rock solid timing or feel of the MPC doing it that way.

Thanks for your input, particularly yours Desmond, it doesnt sound like there is an answer beyond low buffer settings and pulling the audio back manually, neither of which do the job to a suitable standard. This is not an unreasonable thing to want to achieve, it can be done easily in other sequencers and in my opinion this is a massive oversight by Apple, though it looks like I will have to just accept thats how it is.

Thanks anyway.

Hey Stink,

I think you'll find desmond usually dishes out good advice about Logic and this time is no exception. I appreciate this doesn't seem to help your desired workflow.

FWIW, I do this all the time with my SP. If you can record your 8 OUTs at once then you can group these and do a group nudge. That way, the timing of the MPC is retained and sync is established with the rest of your tracks.

And yes, it does depend very much on what those other tracks contain and how much strain they are on your system. You can also try muting or freezing events.

Hope this helps.

ken

--------------------
I'm All Ears.

added to all that's been said, it should be noted that ALL USB Midi Interfaces introduce some MIDI timing jitter , and that often it's not actually Logic that's at fault... the same "inconsistency" of timing issues can show up In PT and Cubase as well.... on PC and Mac. it's because of the nature of the USB bus , and the fact that it's usually shared with other devices....

Martin Walker detailed some of this several years ago in one of his PC musician articles.. measuring a jitter time window of some 11ms on some devices... that's a noticeable problem for doing MPC derived drums and such like....

and external samplers and synths don;t actually play the incoming Midi data "instantaneously" either.... there is inherently a small amount of latency for the hardware as well.... by definition , at the very least there's the D/A conversion latency..... never mind any processing delay in actually generating the sound .... and , sadly, each device is slightly different in that respect.../ some may have a playback delay of 1ms, some may have a playback delay of 5ms, and some may vary depending on what's being played and whether there are any FX applied to it.....

add all that together, and you are NEVER going to get absolute precision and sync with external MIDI devices.... period.....


as such, as the others have indicated, if the timing is critical, then manual re-alignment is going to be a fact of life, in ANY DAW...

reading this with interest.
i've just come back to logic after a few years in PC wilderness...
something that occured to me reading this that i don't think has been touched on - i use an emagic AMT8 midi interface which was supposed to make use of Logic's active-midi-timestamp.
does this technology still exist in logic 8? as this is meant to circumvent some of the midi timing issues with hardware.
personally, i'm quite happy with logic's midi timing - you can get just a bit too-fussy about sample accurate timing when 'feel' is far more important.

s.e.

With every mix I've ever done, I've manually adjusted the timing of just about every single track. I usually take one track for a reference, for example the bass, and build the timing of the other instruments around that.

There are so many factors to consider, and some have been mentioned earlier: Timing inaccuracies introduced by USB and Midi devices, hardware samplers and romplers, the variations in sample starting points with various sample libraries, delays caused by different buffer sizes and AD-DA conversion stages... And lastly the timing variations introduced by human players.

The Event Float, at a resolution of nearly 1000 ticks per quarter note (which is pretty high) is the one tool I use constantly, and I don't see a way of getting around it anytime soon.

Quote Samurai Jack:

---

i use an emagic AMT8 midi interface which was supposed to make use of Logic's active-midi-timestamp.
does this technology still exist in logic 8? as this is meant to circumvent some of the midi timing issues with hardware.

---

AMT was never implemented in any OSX version of Logic, for one fundamental reason.

AMT was a time-stamping technology, which adds time stamps to MIDI events for systems that don't natively support this (Atari, Windows, OS9 etc).
This gave some advantages - in the case of the Unitor/AMT8, events on 1 1 1 1 for different ports could be sent at exactly the same time (they were prebuffered) from different ports, rather than serially which would introduces some timing slop.

OSX incorporated MIDI time stamping technology as part of it's MIDI framework, making AMT not necessary. As long as your DAW supports this feature, you can take advantage of similar timing accuracy with all OSX-compliant MIDI-interfaces as AMT provides (it's not exactly the same, but offers similar performance.)

Also, following on from someone further up - USB jitter is much worse than using regular MIDI cables. Also the MIDI data flowing over USB is the same protocol, the way USB is designed introduces more jitter, so often, MIDI event timing over USB is poorer than using regular simple MIDI interfaces.

(Of course, USB bandwidth is higher the regular MIDI, so you should be able to push more events without the serial nature screwing things up too much - but I"m talking about jitter-related effects on timing, which is different.)

As mentioned, in these days of sample accuracy, MIDI timing is very poor. It's also significantly poorer than the old analog CV systems.

Quote desmond:

---

Use low buffer sizes for recording, and/or jiggle stuff around to get the feel you require. Also understand how PDC works and what it does for other timing dependent things. Run some tests if you're not sure.

Sometimes it's very easy to get caught up in the "OMG! My recorded guitar part is 15 ticks late!" when the timing inaccuracy of the notes you play might be *way* bigger than that.

You just need to know how this stuff works and especially the quirks that often aren't really manual topics, and develop a workflow that works for you.

And another reason some of this stuf gets overlooked is that where once MIDI instruments were a required necessaity for making electronic music, nowadays more and more people are junking their MIDI hardware, and a whole new generation of computer-based kids are working without ever having touched another instrument, let alone understanding what MIDI is.

Yes, in an ideal world you shouldn't have to do all this, and everything should be perfect - but the reality is life is one long series of workarounds...

(Or at least it feels like it sometimes...)

---

I find this answer unacceptable. It is irrelevant how old MIDI is or what current fad has people ‘junking their MIDI gear’. There are a multitude of valid and vital uses of MIDI - sophisticated realtime performance with wind instruments; serious MIDI keyboard performing, avant-garde music and dance requiring MIDI, or plain MIDI performances across the spectrum of music and not just kids playing around with software instruments. So MIDI is the current and only viable protocol for connecting the outside and real world of music and performance with the computer, with control in mind.

Hence - all DAW and computer-music manufacturers should have this firmly in mind. Furthermore, the future of computer music is no doubt in advanced tactile control and in this MIDI will play a major part. So there can be no accepting of the current debacle regarding MIDI capability of recording in current DAWs (and especially Logic), and your suggestion that we should live with it is to accept mediocrity and substandard. Instead, why are all musicians not screaming from the tree tops to demand better MIDI which - is a complete possibility. I'd suggest that the real reason why MIDI is so lousy in its current implementation within the likes of Logic is because of replies like this that suggest accepting the status quo and working around it. There's simply no incentive for manufacturers to improve the situation.

Into the bargain, Logic is about the worst DAW of all for MIDI. Firstly, even though it is owned by Apple, it actually bypasses core midi! Secondly, Logic cannot handle more than 16 tracks of MIDI recording at one go - whereas most other DAWS can handle as many discrete channels of MIDI as are available from the attached MIDI interface(s). So it would be impossible, for example, to record 16 MIDI tracks output from say a Korg M3 Karma Groove while also playing along on a separate keyboard (giving a total of 17 MIDI channels) - because Logic cannot distinguish beyond 16 channels. Thsi si just one example of how primitive it is.

Given these utterly ridiculous specs for MIDI in Logic, one could assume that Apple simply haven't looked at their MIDI implementation for years and any rubbish within how Logic manages MIDI persists to this day.

I've had a minimal experience with DP and Protools and both immetiately exhibited vastly superior MIDI recording capabilities. Certainly in Protools there was never, ever any time delays in MIDI recorded tracks. We'd simply play an external PC sample library via MIDI into Protools, then recorded that performance to a Audio Tracks and there were never any timing issues.

Logic used to claim that they were the best MIDI DAW but this was never actually the case. In truth, its utterly lousy for MIDI, so if MIDI recording is important to you then you should look at other DAWS.

Kevin.

Quote Kevin Nolan:

---

I find this answer unacceptable.

---

I assume you actually mean that you wish the reality I depict wasn't actually the case. I agree with you, but I cannot give advice based on how we'd like Logic to work, only on how it actually does at this current time.

Quote Kevin Nolan:

---

There are a multitude of valid and vital uses of MIDI - sophisticated realtime performance with wind instruments; serious MIDI keyboard performing, avant-garde music and dance requiring MIDI, or plain MIDI performances across the spectrum of music and not just kids playing around with software instruments. So MIDI is the current and only viable protocol for connecting the outside and real world of music and performance with the computer, with control in mind.

---

I don't disagree with you.

Quote Kevin Nolan:

---

I'd suggest that the real reason why MIDI is so lousy in its current implementation within the likes of Logic is because of replies like this that suggest accepting the status quo and working around it. There's simply no incentive for manufacturers to improve the situation.

---

Yes, you are correct. My apologies. Instead of offering practical advice to let you continue to work, I should have said "Yes, Logic sucks, you canot make music in it. Go buy another DAW." That's the best advice I should have given in response to the OP.

Quote Kevin Nolan:

---

Into the bargain, Logic is about the worst DAW of all for MIDI. Firstly, even though it is owned by Apple, it actually bypasses core midi!

---

Can you explain what you mean? No software in OSX can talk directly to hardware, it *has* to go through the OS, which means any devices using MIDI must use Core MIDI in order to do so.

Quote Kevin Nolan:

---

Secondly, Logic cannot handle more than 16 tracks of MIDI recording at one go - whereas most other DAWS can handle as many discrete channels of MIDI as are available from the attached MIDI interface(s).

---

Yes, this has been a bottleneck and limitation of the environment design for a long time, and it does affect the group of users who want to record multiple devices. There are workarounds that can work for some people in some circumstances, but bottom line is there is always only 16 independent MIDI channels that hit the sequencer.

Quote Kevin Nolan:

---

Given these utterly ridiculous specs for MIDI in Logic, one could assume that Apple simply haven't looked at their MIDI implementation for years and any rubbish within how Logic manages MIDI persists to this day.

---

I think this is probably true.

Quote Kevin Nolan:

---

I've had a minimal experience with DP and Protools and both immetiately exhibited vastly superior MIDI recording capabilities. Certainly in Protools there was never, ever any time delays in MIDI recorded tracks. We'd simply play an external PC sample library via MIDI into Protools, then recorded that performance to a Audio Tracks and there were never any timing issues.

---

I think you read into my reply that perhaps you thought I was happy with the current situation, and that I didn't care about Logic's behaviour. This is not the case. I was simply reporting on how it is, rather than engaging in a rant about all the things in Logic I'd personally like improved.

Logic is very advanced in some ways, and very backward and abandoned in other ways - and you can say this for most pieces of complex software - they are all brilliant in some ways, and all suck in others. And this will differ for different users who use them in different ways. The trick is to find the ones that best match your needs.

Quote desmond:

---

Yes, you are correct. My apologies. Instead of offering practical advice to let you continue to work, I should have said "Yes, Logic sucks, you canot make music in it. Go buy another DAW." That's the best advice I should have given in response to the OP.

---

'jiggle stuff around'...'run some tests'...'you need to know how this stuff works'...'people are junking their midi hardware'...'life is one ling series of workarounds'...

which bit of your first post was practical advice that would let me continue to work?? Sorry I really do appreciate your time but there was nothing of any help whatsoever. "Yes, Logic sucks, you canot make music in it. Go buy another DAW" would have been infinitely more useful, at least I would have known where I stood.

I appreciate that there are going to be delays in any MIDI setup, (I've known this for a little while...), but the delays caused by my interface and MIDI equipment are known and fixed (depending on certain settings), so should be easily worked around. But Logic makes that impossible by doing something behind the scenes which imparts complete randomness to the timing of the resultant audio. For example the delay manifests itself in Logic as being positive, ie the resulting audio occurs early , so Logic is definitely doing something because any delay caused by A/D convertors etc would surely make the audio late ...not early!


I'm not a beginner by any stretch of the imagination, I just couldnt believe that everyone just accepted that Logic behaved like this and thought there must be a secret trick that I didn't know about, but I can see now there is not so this thread has served its purpose...if a little patronising.

You wont catch me complaining about it, but I needed to know for sure that's how it was before I could proceed to address it, see?

Thanks all.

Wow, my posts must be coming across really badly...

I'm not in a habit of insulting or patronising anybody here, and neither did I for once think you were a clueless newbie as you seem to think I do. I weigh in to threads that are interesting, or where I can help, or where I can contribute. I'm sorry you found it all so pointless.

> You wont catch me complaining about it, but I needed to know for
> sure that's how it was before I could proceed to address it, see?

So there was value in my contribution, after all?

I'm done here.

with responses like these, it's a miracle Desmond ever bothers to give help and advice.


if you can't be politely grateful that someone is bothered enough to engage you in some sort of dialogue , with the aim of freely assisting you in your endeavors, then that says more about your attitude than I care to.

--------------------
if you don't know who i am, i aint gonna tell you.

there are multiple contributing factors to what you've described.



The existence of Midi Jitter over USB was explained earlier...

and the different latency timings of different MIDI hardware also....


Logic Cannot compensate for these...


the important word here is JITTER>

it means that the timing varies.... by that i mean Plus or Minus.. 11ms is not uncommon....

in that example, this can mean a variance between maximum deviations of 22ms...


if Logic is trying to compensate for a fixed "nominal" transmission delay value , the results can be either early or late for any given note, because of the USB bus jitter...



however, the consistent negative Offset , resulting in early audio, is IMHO, to do with PDC , because this was NOT an existing bug before PDC was introduced.... or if it was, I never noticed it..... but I sure as hell did afterwards....

basically when they did PDC, they failed to Buffer the MIDI data stream by the same amount... as far as I can tell, even when PDC is switched off, there is a differential.... one can theorise for all sorts of reasons why..... but the essential point to realise is that THIS time error, appears to be consistent... for a given set of plug ins....

and that is a crux point.... the overall Plug in delay is determined by the plug ins currently in use , whether or not PDC is active....

For example , Convolution reverbs tend to have longer latency timings... so if you have Space designer, or Altiverb instantiated, you will have a greater offset error than if you have no convolution processing active....


now then..... it's important you realise that there will be a time offset in ANY situation where plug ins are used...

why?

because they require processor cycles to do their work... and this takes time...


Scenario 1) PDC is switched off. Time delay caused by inserting ANY plug ins in the signal path, especially convolution processors, External DSP processors, and so on.... Audio is being sent somewhere, and processed, then brought back in , this takes time..... and with no PDC, you can easily get to a point where tracks are out of time with each other, never mind anything else.... this makes the Audio in the mix late in comparison to the MIDI , which in turn makes the Midi generated audio appear early...

Scenario 2) PDC is switched on . Same time delays exist, but are compensated for in terms of the internal Audio tracks and busses, so that timing remains consistent for AUDIO only tracks.... by delaying tracks and busses etc by the cumulative amount of latency each plug in reports in the appropriate place...
however, the MIDI stream is not delayed , so again, MIDI generated audio appears early....
In a Mixdown situation, PDC can compensate for the differing delays in assorted tracks and busses.... and keep the audio tracks together in time..... but it could only keep it in sync with live inputs by having a time machine and transporting the computer forwards in time to be there before you play the notes.....

Scenario 3) NO plug ins are used at all.... not a single one... and PDC is switched off... MIDI generated audio appears on time ish.... but is still subject to small timing inconsistency brought about by USB bus Midi clock Jitter and MIDI hardware output generation delays. plus the A/D conversion ....

Quote idris y draig:

---

The existence of Midi Jitter over USB was explained earlier...

---

Just to confirm, I am not using any USB, I am not using any plugins, I am using a brand new project, with PDC switched off and buffer settings at 32, with no fixed delay on the MIDI data or audio recording. Your scenario 3 is closest, my problem is that all those delays are known amounts (apart from jitter but i am not using USB) but you still cant compensate for them in Logic because it somehow randomises the timing of the reslutant audio, albeit by a small amount but it still loses the timing of the MPC.

but as already mentioned there is no real solution so I have accepted that.

Thanks again all.

Edited by stinkfinger (26/05/09 11:38 AM)

Quote stinkfinger:

---

Just to confirm, I am not using any USB,

---

now I'm confused.. what are you using as a MIDI interface then?..PCI card?..

--------------------
MBP 17", PC 100(Nubus Protools) Motu 896, X-Station, Logic9, Reason6, Korg legacy, ACE, Alchemy, Emax II, E-Synth, Evolver MEK, Waldorf Pulse and Blofeld, AS Telemark, AS Leipzig-S

Quote Chevytraveller:

---

now I'm confused.. what are you using as a MIDI interface then?..PCI card?..

---

are you being serious?

I think the question is, how do you connect your midi device to your Mac? Are you using a USB-to-MIDI hardware device or are you using the midi port on an audio interface?

--------------------
www.exaviormusic.com
www.manninmusic.com Music Teacher, Isle of Man

I get the question thanks. What I dont get is why he's talking like the only possible way to get MIDI into a computer is via USB? Or why he's made a total noob of himself by trying to insinuate that I have got my facts wrong because I must be using USB to get MIDI into my computer...which I can assure him, I am not.

Gents, my question has been answered in this thread and the concurrent one on Logic Pro Forums, I thank you all.

Quote stinkfinger:

---

I get the question thanks. What I dont get is why he's talking like the only possible way to get MIDI into a computer is via USB? Or why he's made a total noob of himself by trying to insinuate that I have got my facts wrong because I must be using USB to get MIDI into my computer...which I can assure him, I am not.

Gents, my question has been answered in this thread and the concurrent one on Logic Pro Forums, I thank you all.

---

Maybe the timing issue is as to do with the enormous stick up your arse as it to do with Logic.. If you put as much effort into simply answering a question when asked as you do in spouting vitriolic poison you may find you get to a solution quicker..
As it is you seem to be keener on alienating anybody who dares ask a question

--------------------
MBP 17", PC 100(Nubus Protools) Motu 896, X-Station, Logic9, Reason6, Korg legacy, ACE, Alchemy, Emax II, E-Synth, Evolver MEK, Waldorf Pulse and Blofeld, AS Telemark, AS Leipzig-S

Beautifully put....

Word.

--------------------
http://www.christopherferreira.com

Quote stinkfinger:

---

I get the question thanks. What I dont get is why he's talking like the only possible way to get MIDI into a computer is via USB? Or why he's made a total noob of himself by trying to insinuate that I have got my facts wrong because I must be using USB to get MIDI into my computer...which I can assure him, I am not.

Gents, my question has been answered in this thread and the concurrent one on Logic Pro Forums, I thank you all.

---

So how ARE you getting Midi in and out then??


3 common possibilities exist... 2 of which suffer from MIDI timing Jitter....

and the third is increasingly rare.... and I don't recall many multi-port devices... (that would facilitate more than 16 channel MIDI as alluded to earlier.)



and if memory serves me correctly..... SF uses a fireface... over firewire.... and I am given to undertsand that these also suffer MIDI Jitter. (as do , allegedly ALL firewire based interfaces.... although I've not tested every single model.... it is measurable on most. while the Audio is prioritised and time stamped and tightly clock controlled.... it appears that the MIDI is generally a bit of an afterthought..... )

but I don't have one here to test, so the magnitude of that jitter is an unknown factor right now.

however it would explain quite a lot.... (but not his attitude.... anyone else wonder why many of his questions don't garner a huge number of responses......?? )

in terms of Midi CLock Jitter, the old Mac serial modem port was actually better than USB.... although it had less bandwidth.....

progress eh?

Quote idris y draig:

---

in terms of Midi CLock Jitter, the old Mac serial modem port was actually better than USB.... although it had less bandwidth.....

progress eh?

---

The midi protocol should have been updated 10 years ago, but it wasn't. If I remember correctly, there was some talk of "midi 2" for a short while, but it was never implemented.

Quote desmond:

---

OSX incorporated MIDI time stamping technology as part of it's MIDI framework, making AMT not necessary. As long as your DAW supports this feature, you can take advantage of similar timing accuracy with all OSX-compliant MIDI-interfaces as AMT provides (it's not exactly the same, but offers similar performance.)

---

It's good to know OSX has MIDI timing stamping built-in. I've heard this mentioned before - it doesn't often get shouted about, and I'm interested to learn more.

As I understood it, AMT may have helped for various types of slowness or delay inside a computer. But, I understood the timing info was sent all the way to the midi interface hardware, so delays or jitter involving the computer->midi-interface link (USB, or modem serial port before that) did not affect the timing of the final stream that emerges from the "midi out" ports. (think sending single notes at 1 1 1 1 simultaneously to 8 separate midi ports).

Clearly, this requires support from the midi interface hardware, and not just the driver software running on the Mac. Presumably, OSX provides timing info to the Emagic midi interfaces - only it's not called "AMT" now. But what about the other vendors' midi interfaces, with OSX drivers. Is there a list of which have hardware support for OSX midi time-staming ? I'd be surprised if they all did, unless, perhaps it's a part of the USB midi spec itself.

Thanks for any clue!

So instead of replying to the question (an interface name can't be more than a few letters), and now you're found the answer, giving it back to us, you act standoffish and self-righteous?

Not even a bloody link to the Logic Pro forum, FFS...

Quote Tui:

---

The midi protocol should have been updated 10 years ago, but it wasn't. If I remember correctly, there was some talk of "midi 2" for a short while, but it was never implemented.

---

Apparently, "HD MIDI" should be set for release this summer, according to the MMA. Whether this will actually happen, I don't know of course...

Some info:
http://www.midi.org/news/hdmidi.php

http://www.midi.org/aboutus/news.php

Perhaps it might be time for some SOS coverage of this?


Edited by desmond (27/05/09 04:03 PM)

Quote Chris Edwards:

---

It's good to know OSX has MIDI timing stamping built-in. I've heard this mentioned before - it doesn't often get shouted about, and I'm interested to learn more.

As I understood it, AMT may have helped for various types of slowness or delay inside a computer. But, I understood the timing info was sent all the way to the midi interface hardware, so delays or jitter involving the computer->midi-interface link (USB, or modem serial port before that) did not affect the timing of the final stream that emerges from the "midi out" ports. (think sending single notes at 1 1 1 1 simultaneously to 8 separate midi ports).

Clearly, this requires support from the midi interface hardware, and not just the driver software running on the Mac. Presumably, OSX provides timing info to the Emagic midi interfaces - only it's not called "AMT" now. But what about the other vendors' midi interfaces, with OSX drivers. Is there a list of which have hardware support for OSX midi time-staming ? I'd be surprised if they all did, unless, perhaps it's a part of the USB midi spec itself.

---

You'll have to forgive me for not being up on the details. When Logic first went OSX, there was a lot of talk about AMT not being implemented, and why. Back then, we were still fortunate to have Emagic developers online, and they explained that OSX has time stamping built-in, so AMT was not necessary. I recall there was even an article about it on emagic.de, which you might be able to turn up with the wayback machine.

I remember that the implementation differs from AMT - which, as you rightly say, was primarily used to be able to pre-buffer events and fire them at the same exact time from multiple ports.

I remember digging around Apple's developer centre for Core MIDI specs on timestamped events, and while it was briefly mentioned, it didn't go into any great detail. IIRC (could be wrong) the Core MIDI part of OSX was originally designed primarily by the ex programmers of Opcode, with some emagic input, so they were well versed in MIDI and the design of MIDI interfacing and drivers.

My memory tells me that the implementation, if it's taken advantage of, should work with *all* MIDI interfaces - ie it was the driver and the OS (and possibly the app sending the events) that did the clever stuff, so active support from the interface was unnecessary.

But to get any deeper than that, you'd have to speak to Apple devs or anyone writing MIDI drivers - I'm just simply not that geeky!

If you're on the Core Audio dev list, perhaps you could raise it there, someone with more practical knowledge might be able to clarify this. Like I say, most of this comes from my memory of discussions and articles back with Logic started to first go OSX - which was early on (Logic was the first OSX DAW) - probably 10.1 days. You'll have to forgive my lack of detail, as I wasn't even a Mac user back then, so only half paid attention, as it didn't directly relate to me.

So... not sure if any of that helps, but it's the best I can do...

Quote desmond:

---

IIRC (could be wrong) the Core MIDI part of OSX was originally designed primarily by the ex programmers of Opcode, with some emagic input, so they were well versed in MIDI and the design of MIDI interfacing and drivers.

---

Thanks for that.

Yep. We're talking late 1990s, when Gibson (the guitar maker) bought Opcode and promptly closed it down. The fact the main OMS man Doug Wyatt was forced to find new employment may have been a blessing in disguise, as he ended up working at Apple, on the Core MIDI (and I guess Core Audio) aspects of OSX, which was very much in it's infancy then.

Quote:

---

My memory tells me that the implementation, if it's taken advantage of, should work with *all* MIDI interfaces - ie it was the driver and the OS (and possibly the app sending the events) that did the clever stuff, so active support from the interface was unnecessary.

---

That's the bit I don't quite get. Yes, clever software may help deal with slowness and/or delays inside the computer. But if congestion or jitter on the computer -> midi-interface connection is affecting the timing, then surely the only way midi timestaming can help is if the timestamps are sent to, and understood by, the interface hardware itself.

But perhaps USB is fast enough that it simply doesn't matter now.

Quote:

---

If you're on the Core Audio dev list, perhaps you could raise it there, someone with more practical knowledge might be able to clarify this.

---

Good thinking - I'll check that out.

Quote Chris Edwards:

---

That's the bit I don't quite get. Yes, clever software may help deal with slowness and/or delays inside the computer. But if congestion or jittery on the computer -> midi-interface connection is affecting the timing, then surely the only way midi timestaming can help is if the timestamps are sent to, and understood by, the interface hardware itself.

---

I'm not sure - and I'd be interested to find out the implementation as it's come up before - so if you do pursue it further, do post back with your findings - I'd like to fill my knowledge gap here.

If the interfaces are designed and built and have Mac OSX drivers, then it's conceivable that interfaces do have this support built in. Dunno - I'm purely speculating now.

I'll see if I can dig out the Apple dev notes as well...

Quote Chris Edwards:

---

But perhaps USB is fast enough that it simply doesn't matter now.

---

USB is fast enough, and has high enough bandwidth to send nice fat MIDI data streams, but as I understand it it's jitter is worse than straight MIDI, as it really wasn't designed for timing critical data.

Now, if the way the timestamping works can handle the USB jitter and deliver buffered events to the USB<->MIDI interface, it might be less of a problem (not for USB-only standalone devices, of course.)

Checking out archives of the CoreAudio dev list. Found a few relevant bits - will post shortly.

This is how AMT is implemented:

"Active MIDI Transmission is an Emagic innovation that provides a measurable, and more importantly, audible improvement of MIDI timing. Here's how we achieved this new level of timing precision:

A segment of the sequencing engine of AMT compatible software, (e.g. Logic Audio), is put into the MIDI interface, (AMT8 or Unitor8 MkII), where it takes care of sending the MIDI events to the individual outputs with the highest possible timing accuracy.

At the same time, the transfer of data from the computer to the interface is optimized with extreme efficiency by sending notes in parcels during the pauses between the musical events. These parcels are then unpacked in the interface and sent to the individual outputs at precisely the required time. Discover for yourself why the audible difference in timing has gotten the professional world so excited about AMT technology"

Quote desmond:

---

This is how AMT is implemented:

---

Good find. Presumably, that's from OS9 days, when much fuss was made. Back in the day, as well as Emagic's AMT, other vendors (Steinberg and I think motu) had their own versions.

But I've been wondering where all this went with OSX.

As you suggested, checked out the CoreAudio dev list (thanks for the tip). So far, it looks like:

- OSX Core MIDI natively incorporates midi timestamps

- The USB-midi spec does not. Hence timestamps can't reach an interface using generic "class" driver.

- Some interfaces support timestamps (in hardware) and come with a custom driver such that OSX can take advantage. The protocol over the USB link is (due to prev point above) proprietary in nature.

- Timestamping can apply to midi input when recording, as well as midi output when playing back.

- For output, the key property, as supported by certain hardware, is known as schedule-ahead

- Haven't found a list of what hardware supports this AND has suitable drivers.

- Looks like some MOTU 8 port interfaces may do.

- There's an edirol interface with a physical switch on the back to choose between standard USB midi "class" operation (no need to install driver), or, using a proprietary driver, which offers some advantages (timestamping?).

The above is what I've managed to gleam, mostly from the following posts:


http://lists.apple.com/archives/coreaudio-api/2005/Jul/msg00096.html < br />
http://lists.apple.com/archives/coreaudio-api/2005/Aug/msg00076.html < br />
http://lists.apple.com/archives/coreaudio-api/2005/Aug/msg00077.html < br />