Page 8 of 21
Posted: Sat Feb 24, 2007 1:24 am
by iain.morland
Hi,
After a brief break from music-making I have come back to Live (5.2.2) and this MIDI delay issue is confusing me all over again.
I have been running some tests. Record quantise is switched off.
- I record a VSTi MIDI instrument peformance on track 1.
Whilst playing the VSTi, I record on track 2 the audio output of track 1.
This gives me 1 MIDI track and 1 audio track of the same performance.
I then record to audio on track 3 the MIDI playback of track 1.
And here is the problem: the audio file on track 3 is out of time with the audio file on track 2.
In other words, the audio recorded during the MIDI performance and the audio recorded during the MIDI playback are different.
Specifically I have to enter a negative value in the track delay for track 2 in order for it to play in time with the audio recorded on track 3 of the MIDI playback.
The negative value is -1.10ms. I do not understand where this value comes from. It is not the same as the latency setting in the options panel, which is 1.46ms, nor is it the same as my ASIO buffer, which is about 2.9ms.
So what's going on?
Do all non-quantised MIDI tracks need to be time-adjusted?

Posted: Sat Feb 24, 2007 5:02 pm
by iain.morland
I've done more testing and am now even more confused.
Consider the attached screenshot:
MIDI tracks 1 and 2 were recorded at the same time. On track 1 monitoring was switched
ON. Track 2's monitoring was switched
OFF.
As you see the MIDI on each track is differently placed.
But it gets even more bizarre:
Audio tracks 3 and 4 were recorded from the MIDI playback of tracks 1 and 2, respectively.
As you see the resultant audio files are in relatively correct positions...but both are
ahead of the MIDI files!
Note: none of the tracks are delay-compensated, but plugin delay compensation is enabled.
I am at a loss as to what is the right or wrong way to record MIDI or audio from MIDI using VSTis.
Posted: Sat Feb 24, 2007 5:20 pm
by Amaury
iain.morland wrote:I've done more testing and am now even more confused.
Consider the attached screenshot:
MIDI tracks 1 and 2 were recorded at the same time. On track 1 monitoring was switched
ON. Track 2's monitoring was switched
OFF.
As you see the MIDI on each track is differently placed.
But it gets even more bizarre:
Audio tracks 3 and 4 were recorded from the MIDI playback of tracks 1 and 2, respectively.
As you see the resultant audio files are in relatively correct positions...but both are
ahead of the MIDI files!
Note: none of the tracks are delay-compensated, but plugin delay compensation is enabled.
I am at a loss as to what is the right or wrong way to record MIDI or audio from MIDI using VSTis.
Hi Ian,
Not that I don't want to believe your findings, but it makes it hard to compare the two waveforms because of the low amplitude. Would be nice to do the same with very sharp and fast attack sounds.
Question: what device did you use to play the sounds? What method of recording did you use? Resample?
Just to keep people updated: we are actively working on the MIDI timing, and hopefully will come with figures, explanation - facts - and probably improvements in the future, depending on our findings.
Kind regards,
Amaury
Posted: Sat Feb 24, 2007 8:04 pm
by iain.morland
Hi,
In answer to your questions:
* I am inputting MIDI using a Radium USB Keystation.
* The audio is recorded by specifying the input of each audio track as "MIDI 1".
Ok, fair point about the low amplitude. I've tried again, using simpler's "monobass" preset:

In this image the audio recordings of 1 & 2 on 3 & 4 are
relatively correct.
But trying again with the monobass preset reveals the problem described in my previous post.
Recording the audio
playback from a MIDI track
regardless of the monitor settings on either source or destination track creates audio that is differently timed to the audio recorded during a MIDI
performance.
For example, track 3 was recorded while performing track 1 (with track 1 monitor switched on), and track 4 was recorded while playing back track 1:

As you see the version recorded during playback is earlier than the version recorded during performance.
Posted: Fri Mar 02, 2007 2:20 pm
by iain.morland
Bump.
Any more news on this?
As my screenshots show, there's a substantial timing difference between recording VSTi MIDI with monitoring on or off.
This isn't necessarily a problem in itself - all I want to know is which is right.

I can't get my head around it!
Posted: Mon Mar 05, 2007 11:12 pm
by xxxmorphicxxx
Is this the same bug that makes Reason record improperly (delayed) while ReWiring?
Posted: Tue Mar 06, 2007 4:42 am
by Amaury
iain.morland wrote:Bump.
Any more news on this?
As my screenshots show, there's a substantial timing difference between recording VSTi MIDI with monitoring on or off.
This isn't necessarily a problem in itself - all I want to know is which is right.

I can't get my head around it!
Hi,
I don't have a quick answer for that. But I'm there and your findings don't remain un-noticed.
Regards,
Amaury
Posted: Tue Mar 06, 2007 4:45 am
by Amaury
xxxmorphicxxx wrote:Is this the same bug that makes Reason record improperly (delayed) while ReWiring?
Hi,
No,, the latency you experience when recording Reason audio output into Live, while that Reason device receives MIDI from Live, is different, and is dependant on the Rewire technology. It is worked on too, but I can't tell if or when we'll get a better behaviour.
Regards,
Amaury
Posted: Mon Mar 19, 2007 12:00 pm
by popslut
Any news?
Posted: Wed Mar 21, 2007 2:14 pm
by citrik
I've been thinking about why this issue might not be notable for some and a critical path issue for others like myself. I think it might have to do with if you use Live as a mixer or if you have another external monitoring path. I have been thinking about the workaround of not using the monitoring, but that's an impossibility if we're mixing in the box, using Live as the mixer with external gear plugged directly into an audio interface.
Any word on if this is going to have a fix in 6.0.6? Just wondering If I should be testing or digging up my old mixer ;)
Thanks,
Citrik
Posted: Mon Mar 26, 2007 8:13 pm
by docjah
HI,
This is a big issue for me.
I'm playing live VSTI along with a band and recording it all.
The timing is sloppy and you can hear the difference between what went in and what comes out.
This really needs to be fixed.
Thanks,
Dave
Posted: Mon Mar 26, 2007 8:23 pm
by iain.morland
I do not understand why it is taking quite so long to get a clear and detailed public response/explanation from Ableton on this issue.
I mean, how difficult can it be to explain how and why your program works in this apparently unusual way?
Please just set out for us exactly what is Live doing, and in what respects its monitoring behaviour it seemingly different from other hosts.
Surely you must already know - you made the software!
And explaining this will be in your commercial interests. For example I have not upgraded from Live 5 to 6 because I am not yet confident that this issue is understood and will be fixed.
Posted: Thu Mar 29, 2007 10:12 am
by elevation1972
I really hope Ableton can fix this midi delay issue by the next update. I am using Battery 3 with Live 6 and this delay issue is really making my music creation process very diffcult. Keeping fingers crossed...
Posted: Fri Mar 30, 2007 4:21 pm
by roby
Speaking of delays in MIDI, please take a look at this bug/bad behavior as well
http://www.ableton.com/forum/viewtopic. ... highlight=
Posted: Mon Apr 02, 2007 9:03 am
by iain.morland
Come on please. Months and months have passed.
If this is never going to be changed or fixed, or at least clearly explained, then just say so!
As I have already stated, this is preventing me from upgrading Live, and is causing me to look into other sequencers as a replacement.
That is a shame, because I would rather stay with Live.
And I can't imagine that having this 8-page thread in your forum is impressing any potential customers who visit this site prior to purchase... for example by linking from
http://www.kvraudio.com/forum/viewtopic.php?p=2455535