I agree with the OP in that it is very desirable to hear the actual sound coming from the virtual instrument that you are using while recording in midi data for sure.

I am away right now so cannot check until I get back but from memory even in Ableton if you have a super high buffer set you get midi latency with virtual instruments. In order to get the fastest response one has to set a very low buffer setting. (Coming from a world of hardware synths this is what I expect and as far as I am concerned it is the best way to be playing stuff in. Not playing early in order to get things back in time.)

It is also only a one click operation to disable every plugin in your system as well. This is easy and quick.

You still have to fine tune the midi offset in order to get midi and audio lining up perfectly. Sounds like the OP may have not done this yet.

Another option is to investigate the low latency monitoring options. On the playback side you can set a high buffer so sessions play back properly and on the input side you can have a super low buffer set for midi recording.

Jemusic wroteI agree with the OP in that it is very desirable to hear the actual sound coming from the virtual instrument that you are using while recording in midi data for sure.
...

Okay. I agree. But the question here is whether or not it's even physically possible. He said the projects are TOO BIG for lower buffer sizes, which makes real time or at least lower latency monitoring impossible—at least without glitches, including those known to be caused by dropout protection.

My workaround—other than taking a plane to Germany and holding the programmers hostage—is to use the onboard instrument sounds rather than the inherently latent VST instruments in order to get a predictable, real time MIDI performance.

Anything after that is out of my hands and requires tweaking, whether it be MIDI timing offsets or quantization.

But that's just me.

It's easy to get a mixdown and put it in another song to track to then bring the track back to the original song.

I wouldn't try and track to a bloated song.

That's what I do. But there are times when I'm also recording keyboard parts in a pretty dense song during the writing stage. I've survived by finding workarounds for a lot of things.

Tacman7 wroteIt's easy to get a mixdown and put it in another song to track to then bring the track back to the original song.

I wouldn't try and track to a bloated song.

Jemusic wroteI agree with the OP in that it is very desirable to hear the actual sound coming from the virtual instrument that you are using while recording in midi data for sure.

I am away right now so cannot check until I get back but from memory even in Ableton if you have a super high buffer set you get midi latency with virtual instruments. In order to get the fastest response one has to set a very low buffer setting. (Coming from a world of hardware synths this is what I expect and as far as I am concerned it is the best way to be playing stuff in. Not playing early in order to get things back in time.)

It is also only a one click operation to disable every plugin in your system as well. This is easy and quick.

You still have to fine tune the midi offset in order to get midi and audio lining up perfectly. Sounds like the OP may have not done this yet.

Another option is to investigate the low latency monitoring options. On the playback side you can set a high buffer so sessions play back properly and on the input side you can have a super low buffer set for midi recording.

The thing is, I totally understand that latency is inevitable at high buffer sizes. Thats not even a problem for me to be honest, because, as I said I need about one second of playing until I can compensate for whatever latency by shifting my playing a bit early.

The real problem is:

-> me playing MIDI early during recording = timing is on point
-> me playing back the recorded MIDI = everything is played back too early

I'll make a video tomorrow to demonstrate the problem and how it is dealt with in every single other DAW

iadaslan wrote-> me playing MIDI early during recording = timing is on point
-> me playing back the recorded MIDI = everything is played back too early

https://forums.presonus.com/viewtopic.php?p=208522#p208522

iadaslan wroteI'll make a video tomorrow to demonstrate the problem and how it is dealt with in every single other DAW

Make sure to disable Constrain Delay Compensation and/or ASIO Latency Compensation when using Cubase or other compensation techniques for other DAW's. Since Studio One, by default will place note messages at the time they are received by external devices. Studio One simply doesn't have an option to dynamically move notes by the amount of total plugin delay.

Additionally: Studio One can mimic lower latency when monitoring VST instruments using LLM for Instruments. It does that by playing all non monitored tracks a fraction early while monitoring. Also it doesn't duplicate > 3ms plugins to the monitored path (comparable result to Constrain Delay Compensation without user threshold). Unfortunately LLM for instrument will bring a lot of other reliability issues, so I personally would not recommend to go down that path.

So when you're recording tracks, you're purposely not playing in time. And then, during playback, those same recorded notes are too far back in time?

Like I said, if you can get a real time monitoring sound while you're playing, you can offset the recording pass or later select all and move the notes back by the amount of the delay.

Playing purposely out of time with a song is something that, for me, would kill the vibe, instantly.

But that's just me.

roland1 wroteSo when you're recording tracks, you're purposely not playing in time. And then, during playback, those same recorded notes are too far back in time?

Like I said, if you can get a real time monitoring sound while you're playing, you can offset the recording pass or later select all and move the notes back by the amount of the delay.

Playing purposely out of time with a song is something that, for me, would kill the vibe, instantly.

But that's just me.

Yeah well it's not a thing I've ever put much thought to until now. You just get used to the current latency and you simply adjust your playing automatically after you get a feel for it.

So in conclusion it is safe to say that S1 simply never is precise when it comes to midi recording right? Because even the smallest buffer setting still introduces a certain latency by which the recorded notes will be shifted early by S1. So the only difference between say 2048 samples and 32 samples will be the severity of the effect I described.

On my side, S1 does a great job at MIDI recording. The instruments I use, my buffer settings, and when I want to use low latency monitoring are all good tools for my work flow.
That is not to say the problem is only yours.
Just want to state that many end users do not have the issue you describe.
Niles offered some good points for you.
For me, I would never want to "play ahead" in order to compensate for MIDI timing. That would kill the feel in a heartbeat. But to each their own. No wrong or right here.
I just never had an issue with what you are describing on my system.

Just in case this wasn't covered...

If you have dropout protection set to anything BUT Minimum, and you don't have "Enable low latency monitoring for instruments" enabled, you'll get exactly what the OP is experiencing.

iadaslan wroteYeah well it's not a thing I've ever put much thought to until now. You just get used to the current latency and you simply adjust your playing automatically after you get a feel for it.
...

I asked this before and to me it is relevant: WHY are you switching between DAWs so much, and why are you leaving a DAW that works for you?

I'm just curious, because I'm on Studio One because I switched over at version 2 and I was never happy with the GUI/UI of the other DAWs I used (small lettering/two level hidden menus, etc.)

So yeah, I'm just curious. If Cubase works for you, why are you using Studio One?

Thanks for also reminding me what to look out for next time my V-DRUMS are off by a mile. I think I set my DP to medium recently. Maybe that's why it became unreliable when I used them last.

j0001s wroteJust in case this wasn't covered...

If you have dropout protection set to anything BUT Minimum, and you don't have "Enable low latency monitoring for instruments" enabled, you'll get exactly what the OP is experiencing.

iadaslan wroteSo in conclusion it is safe to say that S1 simply never is precise when it comes to midi recording right? Because even the smallest buffer setting still introduces a certain latency by which the recorded notes will be shifted early by S1. So the only difference between say 2048 samples and 32 samples will be the severity of the effect I described.

Not true. Studio One performs well within MIDI specification when it come to timing and notes aren't shifted anywhere unless an offset is used.

roland1 wrote

iadaslan wroteYeah well it's not a thing I've ever put much thought to until now. You just get used to the current latency and you simply adjust your playing automatically after you get a feel for it.
...

I asked this before and to me it is relevant: WHY are you switching between DAWs so much, and why are you leaving a DAW that works for you?

I'm just curious, because I'm on Studio One because I switched over at version 2 and I was never happy with the GUI/UI of the other DAWs I used (small lettering/two level hidden menus, etc.)

So yeah, I'm just curious. If Cubase works for you, why are you using Studio One?

Cubase works very well 99% of the time. The thing that bugs me there is the bussing system and send-fx. Since S1 is made by former Cubase ppl I thought I might wanna check it out.

The thing I like about Studio One is that it has a more visual approach with bussing. You can sort tracks in folders and make those folders busses. Also adding sends or parallel fx is way more convenient and organised.

Anyhow, I really appreciate all the contributions from every one here on this forum - but S1 is not capable of dealing with midi in a logical way so I'm off to a better solution.

Thanks again!

I am running a Focusrite Clarett interface over a true thunderbolt connection (on my Mac setup) and getting 2.2 mS of round trip latency. No need to compensate for any timing there. It is fast and feels totally like a hardware synth. Notes are recorded and played back perfectly for me. Even in large sessions. In fact Studio One's midi timing is the reason I left Sonar which was nowhere as tight.

I often play parts to a click and not quantise and I can do this very well and I love the fact I get back exactly what I played in. I get very close to the same latency figures on my RME setup on Win 7. I had to fine tune the midi offset settings though in order to get this so tight.

I also run a large hardware setup over a Midex 8 (USB card in PCI slot) midi interface and get absolutely flawless timing over that as well. That side of it is totally independent to any audio buffer settings as well for me.

When set up correctly Studio One's midi performance is as good as anything I have seen elsewhere. I also use Ableton in a live situation and get the same excellent performance there too. Studio One is no different for me though.

Exact same experience for me, I love studio one but this is a very big problem! make me thing about leaving this soft I love :

You hear what you record at one spot in time with the other tracks already recorded, then when you playback the recorded midi track it is NOT at the same spot!!
This should be corrected ASAP by presonus team!
No matter which buffer size you use or how heavy is your system the note Should definitely be played back at the EXACT same spot you heard them while recording! if there´s latency due to overload -> we understand that and there's no problem we can adapt but then the other tracks and the recorded track should be synchronized EXACTLY the same way while play back and it is definitely NOT the case.
with my sound engineer we determined in our case the track performed live is around 25 to 30ms too early, compared to what was heard while recording which is in my humble opinion about twice as much as what you need to kill the groove!
Of course then we applied a 25 ms latency but my ears are not as precised as what your expertise and ingeniering should provide. If it’s 26ms then we expect studio one to provide exactly 26 ms so we can trust the software and just try to perform as groovy as possible and this should adapt to any buffer size.
I have to admit it is kind of a no go.
please update the soft,
best
Julien

As for signature,

Macbook pro
I7, 2,6Ghz 6core
Mac os catalina
Studio one 5.1.2 62686
Prism sound lyra2

Reading through this thread I’m a bit confused.

OP writes: “I understand latency so I play my notes early and now they land early.” Then don’t play early because apparently S1 compensates for latency.
“But it sounds right when I play it but not during playback.” Then maybe you’re overcompensating, possibly from experience with other DAWs. You can either get used to working with S1 or you can delay the signal of whatever you’re playing to (but not your keyboard audio). Or you can simply delay the recorded keyboard track a little, to get things back in line after the recording is done. There’s a setting for it in the track inspector.