PT7 and midi timing?

[nikki-k]

Hey Eugene, am I to understand that using a VST version of an RTAS instrument (kontakt) with the fxpansion adapter will result in much more accurate playing? I think their release of v.2 is pending and will support PT7. Thanks.

There are many reports of reduced MIDI playback latency with wrapped vs RTAS versions of the same instruments (NI seem to be the most affected). Some report better performance, better timing, etc. In my experience, it varies a bit. But, the VST2RTAS wrapper is well worth it irregardless...IMO.

[DigiTechSupt]

I wanted to get back to everyone about this thread and revisit a few things.

First, the latest cs4 update for Pro Tools should address some RTAS instrument issues, especially regarding printing to an audio track.

Some of the other timing issues are most likely related to older plug-ins that do not support MIDI time-stamping. If you're using an older plug-in or one that was built using a pre Pro Tools 6.9 SDK, then you will need to contact the manufacturer to ask them to update their plug-in to support MIDI time stamping. This is one reason why you can see improvements using a wrapped version of a plug-in. If the wrapped version of the plug-in supports time-stamping, it will have better MIDI timing than one that doesn't.

Regarding the MIDI timing issues that Eugene mentioned - this is a complicated issue and I want to try to address this as best I can. The methodology being used does not necessarily point to an issue with Pro Tools. I have done numerous tests of the above scenario - one computer with a single track of 16th notes at 120bpm connected to another system. I can record MIDI into a very heavily laden session and I do not see the symptoms being described, outside of the usual MIDI slop that is inherent with a serial protocol. When doing the test yourself, are both systems locked to the same sample clock? Which MIDI interfaces were you using?

My test was with a Digi002R on WinXP going to a HD1 system on Mac. The Mac session had 32 audio tracks, tons of plug-ins and auxes and 3 virtual instruments being triggered by pre-recorded MIDI tracks. I saw no appreciable difference between a single track MIDI session and the fully laden one.

There are several factors that could be at play here. If the two systems are not sample locked, the drift between the two systems can show up as MIDI timing inaccuracies. This is the most likely scenario. Other factors could be the MIDI interface itself, it's drivers and it's interaction with Pro Tools. My test system used the MIDI outs of a 002R and a Midisport 4x4 on the Mac system. I imagine MIDI timing would get even better were a MIDI interface like the MIDI IO, which supports MIDI time stamping, being used.

I hope this info helps in some way.

[Eugene]

Hey Eugene, am I to understand that using a VST version of an RTAS instrument (kontakt) with the fxpansion adapter will result in much more accurate playing? I think their release of v.2 is pending and will support PT7. Thanks.

I can`t tell too much difference in field of timing. To me, wrapped versions are more stabile so I prefer them to native RTAS. And Nikki is right...as always
As for timing issue.. It looks like we`ve messed a bit.. I wasn`t talking about transfering some pre-recorded midi data from another sequencer via midi interface. I experience all symptoms discribed by Jules when I try to record midi - just playing the keyboard. No matter which instrument - RTAS of ReWire (Reason, DFH). Everything`s fine when I`m just playing, not recording, respond is right in time ( I dont take into account buffer latency). Now press record and instrument responds randomly delayed, no matter RTAS or Rewire. So I dont think it depends on RTAS updates.
I don`t think it matters, but I have most popular system for tdm based on ASUS P4C800e deluxe, 1.5gb of memory, 3.2 cpu; Steinberg`s Midex8 with latest drivers. Never had a problem with it in other programms.

[Trailerman]

I should just clarify a few things (from my own perspective). Forgive the use of caps and bold - I'm just trying to make sure the pertinent points are clear.

RTAS instruments/plugin have NOTHING to do with the MIDI timing problems I personally experience and have reported on repeatedly. I can reproduce them 100% reliably in a ProTools session which contains no plugins, or virtual instruments. The issue relates to how ProTools RECORDS incoming MIDI data, not to how it routes it to instruments, or plays that MIDI data back.

To repeat - in a simple ProTools session (just a couple of tracks), PT records incoming MIDI perfectly. Whether it comes from another sequencer or keyboard makes no difference - PT just sees it as incoming MIDI. To run tests, I use another sequencer, because I can quantize the notes, which makes testing the accuracy of what ProTools has recorded easier. And in a simple session, what PT records is perfectly accurate (within the confines of normal MIDI timing drift).

In a more complex session (not crazy - say 64 tracks plus some busses), PT is incapable of recording MIDI data accurately, as described in my post above. It result in late notes, notes held too long or truncated, etc.

The RTAS/virtual instrument thing is a side-issue, and has nothing to do with the issue I am referring to.

And for the record, I think this issue needs to be addressed as a matter of priority. I've been using PT for just under 2 years, and I cannot use it either to host instruments which are triggered from an external source, nor can I rely on it to accurately capture a MIDI performance of any kind.

Jules

[nikki-k]

Jules:
I have not personally experienced this...but, I have not attempted to recreate the scenario you have outlined. After reading your most recent post, it suddenly dawned on me: Why not try a test of your own, that mimicks yours as closely as I can (without making me work too hard hehehe!)

I will try this today:
I have my Kurzweil 2500XS, which has it's own sequencer. To my knowledge, it is accurate as MIDI can be (I hope...)
I have Nuendo3 and PT HD on the same machine.
I also have a Digi MIDI I/O, which should give PT an edge.

Going to do a nice 8 track MIDI composition on the Kurzweil, constant play on each track, for 2 minutes. I will use Channel 1 in and out for the Kurzweil, with nothing else using the MIDI I/O. First, I will lock it up to Nuendo3, and have a light session (project) set up, and record. Then, I will create a more intense N3 session (project), and see what happens. And then will do the same two scenarios on PT HD. (both are on same system) Nuendo3 will be using the Digi HD and MIDI I/O hardware as well- only viable hardware on that system.

Give me several hours, we'll see what happens!

[Trailerman]

Hi Nikki,

Thanks for taking the time to test.

I've sent you a few files off-list which may help you out. I'm pretty sure Digi are aware of this, and are looking at addressing it some time, hopefully soon.

Regards

Jules

[nikki-k]

I tried to use Nuendo3, but I am so unfamiliar with the inner workings- I could not get it to synch and such with my Kurzweil. Pro Tools however...like a dream =D

I went ahead and created .mid file, type 1 (multitrack) at 480ppq. 8 tracks. I recorded myself playing nonsense, copy-pasted across the other 7 tracks, and then nudged each subsequent track's material ahead a couple beats, just to add variation. I put the .mid file on disk, imported into the Kurzweil K2500XS, and set it to sync.

Initially, I created a new session in PT, only 8 MIDI tracks, ADC on. Set input to the first 8 channels coming out of my Kurz into a Digi MIDI I/O. Sent sync from PT thru Digi MIDI I/O to Kurz. Rec enable all 8 MIDI tracks. Hit rec/play, Kurz followed as it should. Played/recorded just over 2 minutes. Next, new playlist in each track. Then, I imported all the tracks from the session Jules (trailerman) sent me. Rec enabled the 8 midi tracks once more, and hit rec/play.

Duplicated the 8 MIDI tracks, and set one to first playlist, second to second playlist/pass. I then imported the original .mid file into the track list (drag n drop), which created 8 new MIDI tracks with the original data in place. Dragged each just under it's correspoding track from the *bounce* from the Kurzweil.

I would really like to know what acceptable drift/inaccuracy is. Generally, I am seeing a delay of about 8-20 ticks for each note. Inherent latency, correct? Next, I am finding different note lengths...variations of 1-20 ticks here!

I never really saw a problem, but this is probably because of how I work: I play in real-time to a track; if I feel good about the performance, keep. If not, another try. Just like audio, I will playback what I have recorded, and do a comp if I just cannot get that one, all the way thru take. Or, it might just be a little nudge here, a little velocity magic there...nothing drastic. If I cant play it, I would rather keep practicing the part until I can play it. Or, if doing dance or "industrial" or such, then a bunch of pencil, pencil, copy, paste, pencil, paste, etc, since the genre warrants a more "machine" like sound at the foundation.

Have not had a chance to go track by track yet. A quick glance at all 3 MIDI track 1's via piano roll and event list was enough to see variance that I would be unhappy with. If it is worse than this for live playing into PT, then maybe I am a better keys player than I thought! ;D lol!

(oh- PS- Digi- PLEASE...PLEASE!!! give us multi-track midi event listing, as well as coincident note AND velocity/sustain/cc editing!!! Just take a quick look at any other MIDI app- Logic, Nuendo, Cubase, Sonar...see how they do it? Do the same please! Just like the tempo ruler is now: simply have an EXPANDED midi track view, with even a single CC below the notes view.)
(oh- and fix the MIDI views problems...ugh...please? pretty please???)

[Trailerman]

A quick glance at all 3 MIDI track 1's via piano roll and event list was enough to see variance that I would be unhappy with.

Hey Nikki

Thanks for taking the time to run your tests.

Although I didn't follow you're methodology entirely (sounds like you were much more thorough than me!), I take it from the above comment that you also saw the the MIDI timing issues Eugene and I are suffering. It's good to have it confirmed.

I have screencaps of what the pianoroll view looks like when I just try and record a single track of perfectly quantized, identical length 16ths into a PT session - late notes, notes held too long, truncated notes etc. Not all the time by any means, but at least one 'wrong' note every few bars. The more complex the session, the worse the problem becomes. I'd post the screencap, but I'm too dumb to work out how to post images - in fact I'm not sure the DUC allows it. This is very different to the inherent drift of MIDI, which will result in a very small divergence (I'm talking a few ticks only) in note starts/lengths.

Incidentally, I totally second your request for more advanced MIDI editing Nikki. It's the only reason I still do most of my arranging in Logic, and use PT for mixing (for which it totally rules).

Thanks again.

Jules

[b.forth]

I sort of stumbled acrossed this thread, and it is of great interest to me as well. Nikki answered a question I had over at Toontracks about Protools and Logic in regards to recording drums MIDI. Thanks Nikki.

I've been over to a friends house who has PT LE 6.9. We recorded my Vdrums TD10 Exp. into Protools as MIDI using a M-audio MIDI device that had the MIDI connectors on one end and USB on the other. We plugged the USB straight into the computer and the MIDI I/O into the Vdrums. The goal was to record the MIDI data of the drums and use Battery 2 as the actual drum sounds. What a nightmare! All the drums on playback sounded like they were lagging behind.

We tried some other things like recording the MIDI data playing things like a double stroke roll on the snare, hi-hats, as well as doing quads (hitting a tom, a second tom, kick drum, kick drum all together really fast). It didn't even track half of the hits. It sounded like a train wreck! I don't know if this had anything to with the Vdrums, the MIDI connectors, or Protools. He just recently upgraded to 7.0 within the last week.

I am less than enthused about this finding with Protools as I am looking for a recording app. to buy, and it has boiled down to Protools and Logic Pro. This is the way I would like to record my drums and use Drums From Hell Superior or BFD as my drum sounds and putting each drum on it own channel. I will however be recording guitar, bass, vocals, and sax as audio. I don't have the resources to record acoustic drums.

If it's that much of a hassle to do MIDI in PT, forget it. I know PT is great for audio though. How is Logic for this type of MIDI recording and audio editing?

Bryan

[nikki-k]

Hi!
Ahh- Bryan- nice to see you over here!

I have a V-Drums kit, with the TD-10. Even tho I cannot play it, I have had drummers over before, and tracked them using it with PT. I have not seen the problems you speak of tho...

I had a couple pads/drums hooked up to it the other day, and was playing DFHS live via them+TD-10. Not a full kit by a long shot, but still had no probs with snare rolls and such- everything I tracke dplayed back/recorded just as I played it. The inconsistencies I found with my system were...interesting. I am still trying to figure a way to play live, and capture the performance in two places at once, reliably, with predictable, expected error. Possibly, set my Kurzweil up to record, sync to PT, and then record in the Kurz while recording in PT. Export the Kurz .mid, and then import into PT and compare. Then repeat with a larger PT sessions running.

The big "test" I want to try is with Nuendo as well, to find out how "inaccurate" it is...