Midi record timing still not right in 5.0.3b4

[amo]

It's been disturbingly quiet here for a few hours....*tick tok tick tok*...

Ok, I don't want to speak too soon my first impressions since I'm the first one to comment...

....but just within the first few minutes of using it, this instantly "feels" better!

Oddly enough, I think the midi clips I record into a single track a with monitoring set to IN seem even more accurate than the ones I record simultaneously on a second track recorded simultaneously set to OFF. The ones with monitoring OFF are just a tad earlier than the ones set to IN (maybe a tiny amount earlier compared to the mentronome than the actual performance was). Chris J and I had discussed this before...that somehow OFF recorded clips can get recorded earlier than the performance the more that high latency plugs are added to the set. Even with this beta you will notice the monitor OFF clips are a hair earlier than monitor IN. I can't guess why that is...but interesting to note that there still is a difference.

The important thing is, I think the clips recorded with monitor IN sound pretty much how I played them compared to the metronome when I play back the clips. I haven't used it long enough to tell if it's perfect but it "feels" pretty good.



I'm going to spend some more time fiddling with it and keep my fingers crossed, but so far so good.

Good to hear that ! but let's more testers come. I'm off midi recording for now, but I will see if I can do some testing.
Cheers,
amo

[diverdee]

sounds cool - I may actually have to grab this beta & give it a shot soon (pissed as a rat at the moment).

[Johnisfaster]

I did a test of recording on 3 tracks all at the same time using each of the 3 methods of monitoring and I did see that in and auto produced different results than in off mode. but it was extremely minute. BUT everyone should be happy to know that if you simply turn off the delay compensation then all 3 methods of recording produce identical results.

anyone else have experiences with the beta yet?

[iain.morland]

BUT everyone should be happy to know that if you simply turn off the delay compensation then all 3 methods of recording produce identical results.

Sorry to ask the obvious, but do you mean reduce the delay compensation (in the "x.xx ms + x.xx ms =" options area) to 0.00, or are you talking about switching off PDC?

I'll try the beta as soon as I can - as I mentioned above I'm a new user, and am still setting up 5.0.2.

[djsynchro]

Boys & girls,

So I tried to do a little bit of "scientific" testing. I have an Ensoniq keyboard with a built-in hardware sequencer. I programmed a simple 16th note sequence on this.

I programmed a drum pattern on a Midi track with Drumular in Live, I got the Ensoniq to send Midi notes to another Midi track with a Simpler bass sound with a well defined attack.

Live was NOT synced to the Ensoniq of course, both were free running.

A lot of hardware sequencers don't play at their displayed BPM so I checked first that when I manually started the two together they would stay "in sync" to my ears for well over a minute.

My qualifications for doing this: I DJ'd with vinyl for a long time & I can play 3 decks so I have a pretty good ear for things being in sync even if I say so myself.

So what I then did was, start Live and the Ensoniq and then fiddle until the two were playing completely tight to my ears, and then record midi clips, both with monitoring set to Auto and In.

I did this with a bunch of plugins: Compressors, resonator, saturator and a return channel with Live's reverb and I also tested it a couple of times with Voxengo Pristine Space with its latency set at 512 samples, and with SIR (this is a freeware convolution reverb that introduces a whopping 8192 samples delay)

Recording this way, Midi clips would be off by never more than 1/128th note (which is about 3.25ms at my test tempo of 144 BPM.) Sometimes they would be off by less than 1/256th.

Can a you still hear 3.25 ms? This will vary from person to person. The ear is very sensitive to timing differences so I did another test and found that if I have two quantised clips playing in Live, (the drum part and the recorded sequence from the Ensoniq) and I nudge the sequence by 128th of a note I can hear a slight difference.

And now the most important thing: After recording I played back the Midi clips, without quantising, together with the drum track to see if what I recorded SOUNDS the same as the way I had it playing. To my ears it does.
This is the thing that matters to people who want to record Midi and leave the recorded performance unquantised, to retain the "feel".

Conclusion: To me it seems that recorded Midi plays back pretty much as it was played now! The notes are off by a tiny amount but it's impossible to tell if that's how Live records Midi or if it's the way I had synced up the two sequencers by ear.



Anyone else did any testing? What are your impressions?

[EDIT] just tried with Simpler playing the drums recorded midi is off less than 1/256th of a note!

[dragonbyte]

Hey everyone !! ...I'd like to jump in on this topic !!!

This one is near and dear to my heart because I track MIDI (unquantized) from a drumKAT midi controller into DFH Superior drummer through Live 5.

After a few of the suggested test you guys have posted, I've found that the Beta indeed records more closely what was played from the drumKAT. I tried recording 2 MIDI tracks at the same time. 1 in Auto and the other Off.... it does a better job for of recording notes and CC data for Hi-hat. The release version would sometimes get sluggish and bogg down a bit when recording notes and CC data .... and seemed to delay the MIDI notes being recorded ..

on playback I did find that the track that was recorded with monitor off indeed recorded a bit faster ..... Roughly 2.97 ms. I arrived at this conclusion by playing back both midi tracks at the sametime (same exact sample). I heard a very slight bit of flamming so I offset the Track (Set to off) untill I could get as close to phase canceling as possible ... and thats how I got 2.97ms which is interesting for me because thats only 2 100ths of a milisecond off from 25% of my overall latencey :/

Anyway after sharing all that mumbo jumbo I think the timing changes in the Beta are a plus for incoming MIDI recordings (for me at least ....) I do still offset DFH's MIDI track after recording to roughly -5.00 ms I believe that this is just it's built in latencey window. Thats taking into account the periceved attack of the samples.


Best Regards,


Chris

[dragonbyte]

Just tried Beta 5.0.3b5 ....

It closed the gap of 2.97ms for a track with monitor off/Auto to 0.13 ms ...
Thats pretty amazing !!! ....

well done guys ..... : )

-Chris

[Nod]

Just tried Beta 5.0.3b5 ....

It closed the gap of 2.97ms for a track with monitor off/Auto to 0.13 ms ...
Thats pretty amazing !!! ....

well done guys ..... : )

-Chris

Seconded - just posted a few quick impressions to the top sticky - and again thanks to everyone on this thread. This really was/is/and will be of crucial importance when it comes to Live's development as a fully fledged studio tool.

The whole paradigm of a sequencing instrument was somewhat spoiled when every MIDI/VSTi performance had to be corrected, by hand, in order to restore the correct 'groove'. There may still be further work required, as we should try and get this as close to perfect as is possible, but from trying out 5.03b5 huge progress has quite obviously been made.

I can now get back to recommending Live to every musician I meet rather than qualifying that opinion with a warning of ropey MIDI timing

[djsynchro]

Alex,

Please reply and confirm that you understand what is needed here and that it is being fixed very soon.

If this doesnt get fixed soon, then I'm selling my license to someone who doesnt need MIDI functionality.

Yo Les!

Wassup boyyyyy?
Like the new Beta?
Like the timing?
Yes? No?

[iain.morland]

Delighted to hear that progress has evidently been made. Kudos to Ableton for the fix.

Regarding that remaining very small delay, can someone tell me (as a newcomer) whether this is something that should be compensated on individual midi tracks (in the 'multitrack' view), or by adjusting the overall midi latency in the options panel?

Thanks

[diverdee]

bumpity bump - haven't tried the beta yet - been out for a beer or three - but the news is very encouraging, hope it continues in that vein & then I can definately put aside some money after crimbo for the 4-5 upgrade.

[negativeinjury]

I also gather (kind of) the theoretical explanation given - but as icedsushi explained also - how come every other application doing this manages to do it seamlessly & tight?
If it's not possible how do they do it?

They don't do it!!!

Cubase has the same problem of bizarre midi offsets since full PDC was introduced in SX2, as does protools, since full PDC was introduced on that sequencer in version 6.4.

Please check this thread, where I am trying, and failing to tell a cubase user that the same problem occurs in ableton, and that it isn't steinberg's fault!

http://forum.cubase.net/phpbb2/viewtopi ... sc&start=0

Midi timing in Cubase is a great deal better than it was for the 3.0 release, but for certain interfaces there are still major offsets to contend with in version 3.1.1..

-Fergal
(Registered cubase user who makes no claim to be a Live user)

[icedsushi]

Hmmm...since this beta, could that mean the Live has some of the best midi timing compared to other sequencers with PDC turned on now?

(I used to be a Logic user. Never tried cubase or DP myself, just never heard anything bad from users about the midi timing.) I had problems with weird midi timing bugs in Logic ever since they put PDC in, lots of stuff was strangely too early, no easy workarounds etc.

I am very happy with this beta. I will admit, happier than I was with Logic 7.1 all-around midi timing. There, I broke down. Even with the occasional drama, I wuv u guyz!

[Nod]

Hmmm...since this beta, could that mean the Live has some of the best midi timing compared to other sequencers with PDC turned on now?

That's what I thought Sushi.... Some folks, who aren't making a lot of use of the MIDI side of things, might not think this is a major bonus. For me personally it's one of the best advancements ever to happen to Live - MIDI sequencing can be a constant day to day job in the studio and anything that relieves it from being a chore etc....

I am very happy with this beta. I will admit, happier than I was with Logic 7.1 all-around midi timing. There, I broke down. Even with the drama, I wuv u guyz!

Welcome to the Live V5 MIDI Counselling & Support Group! Seriously tho' now it means we can actually get back to the genuinely cool stuff - playing and writing

[dragonbyte]

Hey Guys !!!

I'm suprised no body has mentioned MTS midi time stamping in this thread ...
given the topic ...

The timing asdjustments are huge for me in this new Beta and are working great !! ..but I think what we all really want is MIDI time stamping ...

There are only a few MIDI interfaces that do this like DIGI MIDI IO and most of the MOTU interfaces .. And only three DAW apps that will use the timestamping
if the interface transmits it. Protools (with DIGI MIDI IO only) DP4 with all of MOTU'S interfaces and Logic Pro (this is built into Core MIDI from what I understand) i'm not sure about Neundo or Cubase...

I don't know..I could be wrong but, I think this really is the only way to get absolute accurate MIDI timing ...

Having said that we still need to use the PDC to compenate for the latencey of the plugins :/ I don't think there is any way around this unless the makers of the plug admit to their actual latencey and some how tell the Host app (Live) to always offset by X amount to be completly accurate on play back ..

-Chris