Hey Lovers,

I came across an issue in FL Studio leading to the conclusion that FL Studio is not operating sample accurate when editing stuff. I guess the conclusion is wrong and therefore hope someone can help how to resolve my issue. Maybe a feature or a setting can do the trick.

Situation:
Having a drum loop I recorded a decade ago (1 file, stereo, full drum track, 3 min.), I wanted to "replace" the drums. Not on audio Insert FX level as apTrigga3 but with the power of Midi, reprogramming the drum pattern. To easen up things I max. zoomed in looking at the wav drum track above the pattern where I programmed the drum events.

Problem:
The drum loop contains events that are off-grid, swing, bounce, feel, whatever you call it. I am able to come close to the loop with quantized events. But when I turn off all quantization to reposition the slightly missed events, even though having zoomed in max I am not able to position the midi note in the piano roll right below the event in the audio file. The smallest step I get is not enough to get it. It's either too early or too late.

In Cubase I can edit on sample level. Very useful when cutting audio and avoiding crackle. So where should I go in FL Studio to get it done?

 

yea use the new time (left click edit in time warper) plugin put the drumloop in that set the hit slices and then youcan drag each hit to the grid or out ofthe grid youcan also re adjust the groove of the drumloop aswell

 

Last edited: Nov 26, 2020

Go to settings click on project increase timebase ppq to the highest value. Lets you zoom in more. Uses more cpu.
Hold down alt while shifting the midi note.

 

Nice one, Just looked it up, right clicking the "create score" button seems to do my midi magic. Thanks, will try.

This looks promising! Thanks man

 

Update: Got the drum midi event now to where I wanted it to be. So thanks for helping with the issue. (turned timebase ppq to max 960).

But still, when trying to trim the audio on sample accurate level (zoomed in to where you can really see those edgy stairs of the waveform on each sample step), looked like not possible. There might be a workaround maybe through edison or smth.

How do you guys do the trick?: When you have a track in the playlist and you want to cut between Verse and Chorus, and crackle occurs as the waveform is not at zero on the cut you did. (without fading out or any volume automation, which could be recognized as a gate effect or smth)

 

I just check zero crossing and call it a day, I don't know if it really hits zero? or just as close as the ppq/grid goes?
anyway, if that doesn't fix my problem I select the de-click option in the sampler, which are just fades.

I do realize Image-Line should fix this, these are just workarounds.

 

You can only move audio according to the audio resolution, aka sample rate and MIDI events according to the MIDI resolution in PPQ.
Means, at a tempo of 120bpm and a SR of 44,1kHz you'll need at least 368PPQ (44100/120) to move MIDI events with sample accuracy.

 

Flames, thanks for sharing this.

 

For accurate sample editing fire up edison, soundforge gold wave etc. you can move the wave inside the clip increasing timebase while you do that after your done turn it back to default

 

Because you would have to convert the minutes (in bpm) to seconds (in Hz) first, it would be actually 44100*60/120 = 22050.
Thus, you would need not only at least 22050 ppq, but exact 22050 ppq or a multiple of it. That's why sample accurate MIDI is impossible in general.

 

Math for noobs, you're right.

EDIT: Now here's the funny thing:



Maximum zoom, you can see the sample steps in the first track.
Second and third tracks are MIDI, with the third just being a copy of the second.
The two black lines are MIDI notes.
The gap between these notes is the smallest amount by which I could move the lower note.
But, as you can see, I could move item (part) in the third track for less than a sample .

 

Last edited: Feb 2, 2021

Now this is just LUDACRIS!!! Editing within the MIDI you get an accuracy like 40 times worse than when you move the event itself? Now if this was science, would you print both to wav and clear wheather you have indeed raised the accuracy for that one note or if it just cheats you by showing a difference but when the VST gets the info to make a sound, just treats both identically (output identical).

Standing ovation for putting that together, I somehow avoid my initial problem here with the tips of zero-crossing and the options in the sample editor in FL. I guess my initial approach (as a Cubase longtimer) was to take the same action known but all that counts is the result.

 

Ok, that's strange.
But I guess it's only in the view. If FL Studio can show you the raw values somehow (like the list view in Cakewalk or Cubase), you could check if there's really an offset between both clips...

What timebase do you have in settings? AFAIR as with most other software, 960 ppq is still the max. there. (I don't have FL Studio right now...)

 

Nope, it works, to my surprise - and confusion.

I rendered the MIDI note (short click), phase inverted the audio -> flat line, complete canceled out. Then I moved the item by a tiny bit more than 1 sample and this is the result



Visible and audible. They are not canceled out anylonger. Moving less than a sample keeps the output dead silent.

960PPQ

 

Nice work. I am confused, still.

Isn't that proof of only showing you a better accuracy but not delivering it?

 

This has been known since the start of daws! I'm surprised you think it's ludicrous.
This is why myself and my colleague would spend hours bouncing down stems in Logic and reimporting them and shifting them! In 2003!
And imo even in the present day one should still always bounce all MIDI and then still do some dragging accordingly.
It sounds better, always.
wow.

 

IMHO, not. The audio output just doesn't change when you move the note by less than 1 sample because the audio resolution is discrete, not continuous.

 

You should better not be surprised about people being surprisingly "uninformed" (you can also call that dumb).

You are talking about the accuracy of MIDI events compared to audio events at a certain sample rate. But this was about MIDI events vs MIDI events that have been moved as @No Avenger pictured.

What sense would it make to have 50 different positions for MIDI information, that all lead to the same resulting audio output? The answer is none. If there was sense to it, then audio events would also be movable by less than one sample - just without any audible or measurable difference in the output. But aight, as this would lead to confusion audio events snap to the grid, so if you move it, it does actually change.

Why should any DAW Maker create potential confusion by showing you stuff that seems possible but is not. Maybe that is for people changing the samplerate afterwards, but do they really already know where to put that information so it can shine even better in the future after changing the samplerate?

Technically totally understandable. But from perspective of usability not.

I think if both outputs are the same the visual accuracy is actually cheating you.
And if both outputs are different, there was reason to show you a difference.

It should very much be possible for a DAW to compensate for the lack of MIDI accuracy under the hood. But usability-wise this could look awful.

 

Last edited: Feb 5, 2021

Yes,agreed. Remember also that the placement of audio is within the same domain of a point in the MIDI world too.

I kow what you are saying, but it doesn't actually work out like that in practice from my experience.

Agreed. It's strange.

I don't think any daw has done this perfectly form the start.
Which is why for 20 years we would not trust the MIDI bounce and move it, and also move other audio around.
It's the difference between a track that pops correctly, or doesn't.
For whatever reason that may be.
I've never totally understood it either.