PT 8.1 CS2 - When Midi Timing becomes a variable !!!

[25ghosts]

Hi,

I am WELL aware of various midi issues in PT like the ADC with Midi etc.

But this one has me going nuts.

Lets say I start a session running @ 128 samples in the H/W buffer.
I like to start off like that to get a minimum midi response (Yes increasing the H/W Buffer WILL also affect the midi delay to external synthesizers.)

So a while down in the composition, I have about 16 midi tracks running - some to external synths some to PT VIs and some to Vienna Ensemble pro. And depending on the plugs I use etc I will typically have to up the buffer to 256 or 512 or even 1024... Every time I do so I sense the groove of the composition changing. But I always thought, naaa - it is the same effect as tweaking a hiShelf EQ (That aint on) hearing how the highs are being added to a snare... But today as I upped the buffer to 2048 I KNEW that I WAS indeed hearing a difference in the groove.

So I decided to a test -

The test is like this:

16 Protools midi tracks playing midi to various sources which are being fed back and recorded onto separate audio tracks in PT which were assigned to be summed to stereo track via bus 1-2. ADC has been set to OFF and no plugs aside from the VIs are sitting in PT.

First to make sure that each midi pass was the same I would recorded the
the track twice set at a buffer of 128k. Swapping the phase on either would in this case null perfectly.
So I kept one recording conducted at 128 at did a new one but this time with an H/W buffer set to 256. The result was that the two results would NOT cancel out perfectly.
Upping the H/W buffer to 512 and comparing that to the 128 buffer would yield even greater differences. The higher the buffer went the bigger the differences.

YES - I DID take into consideration that I had to nudge forward the track by the amount of which the buffer was greater... So all the results and phase cancellation tests were done with the tracks aligned.

This to me is extremely disturbing. I would like my composition to stay the same regardless of buffer size set in the H/W Buffer of the Playback Engine Dialog. Of course, setting the buffer to 256 instead of 128 WILL add an additional delay of 128 Samples and so forth.


I have tried the same in Logic pro but was not able to provoke this weird behavior.


Does anyone have an idea what I could do to alleviate this problem...

I know I could try an stay at the same buffer but then again one has to ask himself - if ALL the buffers are yielding various timings WHICH IS THE CORRECT ONE ---- The one I played.

[kingchong]

While doing a search on Google for MIDI Timing issues in Pro Tools, I came across this thread. Although I am on a completely different system (Win 7 64 bit, PT LE 8.0.4), I am experiencing similar problems.

I always wondered why after I would track my songs out, they would never sound quite right. I started to investigate the Audio Tracks I recorded alongside the MIDI Tracks, and realized that the transients of the Audio was not lining up with the MIDI Note, etc. I messed with the Buffer settings, and using a value of 64 brought it close, it was still off.

I did a small test by creating a Session with a simple drum beat at 120 bpm, Kicks on the 1's and 3's and Snare on the 2's and 4's for 8 Bars. I recorded it 4 times, and on each track, the notes hit at different times. I was trying to get a constant number thinking I could set the MIDI Playback Offset, but the timing differences varied too much to enter a value. I then did the same thing in FL Studio and recorded the output in there, and imported the Audio into PT. The transients were still off grid, but were consistently early by 167 or 211 samples. In PT the drum hits varied from 329 to 485 samples (early) and everywhere in between.

Currently, I have my Fantom X connected to a MOTU MIDI Express 128, which is connected to the computer via USB. I have also tried it with my Fantom connected straight to the computer via USB with the same results.

Tomorrow, I might connect my MIDI from the Fantom directly to the 003 and see if the results are any better/worse.

[GS-DK]

Once again - midi is almost useless in ProTools if you want decent timing. It's no big secret.

[25ghosts]

kingchong,

thanks for chiming in bro'

Started to believe that I was getting seriously mad.

I am doing the following now...

1) I open Vienna Ensemble Pro in stand alone mode (Not server via RTAS)
2) I use my MPC4000 as midi sequencer feeding BPM, MF2 and Kontakt4 along with a few other VIs all inside VEPro
3) I have about 30 Stereo tracks going with large arrangements
4) I have the buffer set to 128 samples on my RME FW800 which feeds a digi 192 i/o digitally.
5) The CPU is maxing at 47%....

And the best of things - I can set my buffer ANYWHERE between 128 & 3072 samples and the timing REMAINS THE DARN SAME.... Now THAT is what I would WISH for in PT8...

[25ghosts]

Originally Posted by GS-DK

Once again - midi is almost useless in ProTools if you want decent timing. It's no big secret.

Still Protools features great midi capabilities. I took the time to get down 'n' Dirty with its midi and found that PT8 took a leap towards midi but if the buffer changes the midi timing what good is it...

Set the buffer 1024 and start playing fast notes like 16th notes or persian articulations on a 95 bpm track - I promise that the it is going to sound like you cant play for SH+++++

I think this should be taken care of in an "Industry Standard" Music creation software.

[GS-DK]

I have given it several tries to make all my music in ProTools and I do like the features even if they are much simpler than Logic midi. I can live with that but being a trained musician I can't live with the timing issue.

[drBill]

Originally Posted by 25ghosts

I think this should be taken care of in an "Industry Standard" Music creation software.

Indeed, after upgrading to two MacPro's to run PT as my main midi sequencer with VEP/Magma Chassis, etc., quite disappointing.

Keep us up to date with your experiments.

bp

[25ghosts]

Originally Posted by GS-DK

I can live with that but being a trained musician I can't live with the timing issue.

I can only agree with that. A real shame that is. I HATE leaving 'tools for doing other stuff. But to me, groove is more important than than ANY DAW.

[nst7]

The midi timing being less accurate at higher buffer settings is actually stated explicitly in the 003 Getting Started guide. At least as far back as 7.3.1.

I know that's LE and this is HD, but it shows it's something they've been aware of for quite a while and are admitting it. I have no idea why they haven't fixed it yet.

[Ramtitam]

Originally Posted by 25ghosts

Still Protools features great midi capabilities. I took the time to get down 'n' Dirty with its midi and found that PT8 took a leap towards midi but if the buffer changes the midi timing what good is it...

Set the buffer 1024 and start playing fast notes like 16th notes or persian articulations on a 95 bpm track - I promise that the it is going to sound like you cant play for SH+++++

I think this should be taken care of in an "Industry Standard" Music creation software.

25 Ghosts, I have a huge template with buses and mix a and mix b. When I have a large session delay I always have to reroute my recording midi channel to the main bus. When I do that, the MIDI timing is not affected by the 1024 delay.

I've also experienced strange delay behaviour in PT though.