Can someone confirm incorrect VSTI midi timing

anti-banausic · Post by **anti-banausic** » Tue May 03, 2005 12:23 am

I don't know exactly where the issue is, but I am pretty sure it is not in resampling (though through this, we can see that the waveform is behind).

I have noticed that in some, or most, drum machines (not impulse), the beat is behind the metronome. This is where it is easiest to tell. More difficult to tell with synths, though some arpeggiators seem to lag slightly behind the beat, so I am guessing that they are not receiving the MIDI perfectly in time?

However, I would say it is workable right now. For all of the good things that Live4 is!

But probably should be addressed. It is slightly a pain to fix the first warp marker after re-sampling. It also causes what should have been a perfectly cut loop to be short, and therefore more mucking is necessary.

PDC would be huge for LIVE. Freeze would be beneficial, but I can re-sample (which I am doing a lot with Minimonsta!). However, it may be difficult to implement PDC in LIVE what with the audio enging being gapless.

We can only wait and see. I am really pretty sure at this point that it won't be addressed in any future revisions of LIVE4. This is going to have to wait for LIVE5.

Happy re-sampling and warp-marking.
AB

ilia · Post by **ilia** » Tue May 03, 2005 12:37 am

ct43 wrote:this a PDC problem as far as I know... its a pain in the ass!

I don't know. A good drum machine or sampler should not have any delay, it should work with the buffer given by the host and trigger a sample at a time specified by a noteOn message. So, this is not a PDC problem. Nor should it depend on the buffer/ASIO/drivers etc. If you are rendering, say, a quantized midi loop playing into Battery, and the samples don't land squarely on the grid, this is a buggy hosting and/or rendering implementation.

And yes, I have noticed the same thing with various plugins and synths that shouldn't incur any delay. I asked an M-audio guy doing a Live presentation in New York about it, and he said "it's been fixed", but that's not the case, unfortunately. It would be nice if one of the Abes commented on this, whether this is indeed a known bug and what is the best workaround.

ct43 · Post by **ct43** » Tue May 03, 2005 11:29 am

I would have thought that seeing as they are VST plugins, they would still incur plug in delay... maybe im wrong, but I cant see why they would be any different from FX plugs.. isnt it still the same thing?

mikemc · Post by **mikemc** » Tue May 03, 2005 8:36 pm

I had experienced this w/microtonic the other night. I fiddled with my prefs and it went away. I think that may have been more related to the audio drivers than anything else, I have not been playing with Live in 'computer science mode' so haven't noted exactly what the issues are, and I had a hardware mishap last night that has rattled me a little. I'll try to recreate what I did if I get a chance.

Using FLStudio, I have noted in passing that if you arm your input track to record and monitor to "Auto", there aren't problems. But using monitor set to "on" without track armed, I've had problems. ALSO, using FLStudio in rewire mode has worked *better* for me than using FLStudio in VSTi Mode.

Maybe someone can propose some experiments/provide some use cases to narrow this down?

nebulae · Post by **nebulae** » Thu May 05, 2005 10:04 pm

I can confirm this with VST instruments. I run an XP-based lappie with RME Cardbus and Multiface. Regardless of my buffer setting, whether its 2ms, 6ms, or 48ms, I get a 3ms delay in the output of any VSTi. This is the case with Battery, Pentagon, etc. I tested this out by recording the output of a VSTi as an rendered file as well as a resampled file - in both instances, there was a 3ms delay of the audio.

Conversely, with Live's Impulse, there was no delay at all, regardless of resampling or rendering.

Although I was concerned at first, I recorded the output of a VSTi for several bars. After going into the clip properties and adjusting the start time by 3ms, I went to the end of the clip and found that the timing was sample accurate. So the workaround is to just time-correct the starting position of a VSTi output.

I forgot to mention that I have done the latency compensation test as stated in the Live manual, and have it set correctly. Therefore, what this test implies to me is that Live must have some unique way of rendering audio from non-Live instruments that makes them sample accurate + internal latency. Could this be the reason why Live is inefficient with VSTi performance?

Performing the same test with the same VSTis in Cubase SX2, I noticed no delay in the audio.

Abletons, can you confirm this and please offer some explanation?

conny · Post by **conny** » Thu May 05, 2005 11:42 pm

Resampled PolyIblit - 3 ms delay.
Resampled Simpler - about 0.1 ms delay
// C

katha · Post by **katha** » Sun May 22, 2005 11:57 am

i do have the same problem and find it very annoying!

could please anyone of the ableton team tell us if they're planning to solve this? and about any workarounds, if there are...

thanks, katha

mr-e · Post by **mr-e** » Sun May 22, 2005 1:32 pm

Hi , i use battery 2 in ableton too and i also noticed that the drum timings didn't sound consistent . It seemed 1 measure played faster/slower than the next one , like the midi clock was flipping.

However , after rendering a battery track to another audio track , I noticed that no matter how many cells i played or how many compressors , filters & envelopes i used , there was always a delay of 5 ms compared to other tracks.

--> i sent all non battery tracks to a simple delay set to 5 ms and now everything runs in sync and the timing inconsistencies seem to be gone too. Imo our brains interprete the whole timing of a song as inconsistant as soon as the drumtracks falls after all the other sounds.
I'd have to test with heavier ableton tracks to be sure though.
And I'm dying to find a way to delay my miditracks to external gear , because they still come 5 ms before all audio and nudging them 1 by 1 is not an options if i still want to edit them later.

anti-banausic · Post by **anti-banausic** » Sun May 22, 2005 7:36 pm

So, I never get on the wonderful people at Ableton about anything, because quite frankly, I love this software, and since I couldn't begin to write the darn thing...

But, it is strange that no one from Ableton has commented on this. It seems like there are plenty of people that are experiencing this, and it is quite a drag. Yes, plenty of workarounds, but Pro software should work, period.

I also am fully aware that they are probably working very hard right now on LIVE5...hopefully. And maybe this will be addressed in LIVE5, but just strange how this hasn't been addressed.

ethios4 · Post by **ethios4** » Sun May 22, 2005 8:22 pm

Its super annoying to have to manually retime everything. If I want to route each drum in Impulse to its own track for FX I have to resample each track, manually line up each clip, then render, and sometimes readjust again. I love Live for its incredible workflow, but this PDC problem really gets in the way of that. So much so that I've been considering using another app for studio work. I'll wait til Live5, but I'd bet Ableton would lose some customers if this isn't fixed in 5.
I think part of the issue is that PDC would interfere in a live performance situation. As Live grows, the number of possible uses for it grows as well, which makes it difficult to satisfy everyone. It seems to me the simple solution would be to make features optional in Preferences, so you can turn off PDC for live use, and turn it on for studio.
I really hope Ableton uses Live5 to perfect whats already there instead of adding a bunch of new stuff. Or do both

anti-banausic · Post by **anti-banausic** » Tue May 24, 2005 6:36 pm

I agree whole-heartedly with this.

If it is going to create a problem with the audio engine in live performances, just make PDC an option that can be turned on/off (if that is possible, hope that wouldn't mean two audio engines).

This is really starting to get irksome. You can definitely get ideas down, and it may be ok if things are a little off. I can "hear" what things are supposed to sound like, but in the end, it makes it difficult to finish off things if you want them to be really on.

Still no word. The beta for 4.1.3 is out, and no word of this being fixed. It may be too large for a fix before LIVE5, but I hope this is a PDC problem and will get fixed with that implementation. If it is an internal routing latency issue, that could be more problematic.

raapie · Post by **raapie** » Wed May 25, 2005 6:03 am

have you set the Overall Latency correctly? read page 60 of manual. essential tweak if you want audio-recording with latency compensation (including ADDA convertions). this is not the complete story, some VSTi's are causing more latency than others. Reaktor 4 is a bit sloppy for example, probably caused by object in object configuration of Reaktor (loading a ens-file into the Reator-shell).

Nikocat · Post by **Nikocat** » Fri May 27, 2005 6:15 am

Yes, it seems the ableton are aware. Here is the reply they gave me:

---------------------------------------------------------------------------
"Hello,

Live does not have a track/VST specific latency compensation yet.
This is the reason for this behaviour.

Best,
Christian Kleine"
---------------------------------------------------------------------------

While i appreciated the reply..... the near-term outlook looks bleak.

But i do have good news!
I just saved a ton of money on car insurance by switching to geico!

olafmol · Post by **olafmol** » Fri May 27, 2005 8:29 am

Nikocat wrote:Yes, it seems the ableton are aware. Here is the reply they gave me:

---------------------------------------------------------------------------
"Hello,

Live does not have a track/VST specific latency compensation yet.
This is the reason for this behaviour.

Best,
Christian Kleine"
---------------------------------------------------------------------------

While i appreciated the reply..... the near-term outlook looks bleak.

But i do have good news!
I just saved a ton of money on car insurance by switching to geico!

hmm this doesn't explain tempo changes over time.....

WaveRider · Post by **WaveRider** » Fri May 27, 2005 3:05 pm

I read this whole thread and find this TERRIBLE. I just bought Live 4 2 weeks ago and were not aware of that. This borders on freaking fraud!!!!!!!! More than 500$ for a music app that can't keep it's timing.

Ableton, this is VERY BAD!!!!!!!!!!

Can someone confirm incorrect VSTI midi timing

..

Re: Can someone confirm incorrect VSTI midi timing