Basic MIDI question

[Phil O]

I'm not a MIDI guy at all so forgive my question if it seems too basic.
I don't have patch numbers, etc. for the keyboard that we'll be using, so what I'd like to do is record a MIDI performance and play it back at mix-down to the keyboard for audio, but just send basic MIDI performance info and select the patches at the keyboard (as opposed to sending patch data). I don't even know if I've explained that right. Does that make sense? If so, how do I do that?

Phil

[stubbsonic]

Well, it kind of looks like you answered your own question.

You should try to record the MIDI tracks while using the final/keeper sounds. It's often weird replacing the sounds you used for recording with different ones you use for mixdown.

If the keyboard can operate in MultiTimbral mode, then you can play different patches on different MIDI channels at the same time. You would use the keyboard's interface to select (page?) to the different channels and then enter the program for each channel. That process would vary between models. Make sure the MIDI tracks don't have any program changes in them, otherwise you can select your patch, and the MIDI track will switch it back to whatever is on the track.

Then as you play back, you can record the output of the keyboard back to audio tracks.

I'd STRONGLY suggest you do multiple passes and record the keyboard playing each MIDI track one at a time into separate audio tracks. Then when you have all those synth tracks recorded as audio, you can do more things like panning and EQ separately on each track.

[Phil O]

If the keyboard can operate in MultiTimbral mode, then you can play different patches on different MIDI channels at the same time. You would use the keyboard's interface to select (page?) to the different channels and then enter the program for each channel. That process would vary between models. Make sure the MIDI tracks don't have any program changes in them, otherwise you can select your patch, and the MIDI track will switch it back to whatever is on the track.

You're assuming I understand anything you said.
The thing is, we'll definitely be selecting sounds at mix-down and don't have the luxury of recording the exact sounds while tracking. I realize it's not ideal, but it's how this is gonna go. Don't ask, I got overruled on this one.

Make sure the MIDI tracks don't have any program changes in them.
How do I do that in DP? Can this be accomplished while tracking?

Sorry, but like I said, I'm not a MIDI guy.

Thanks for the reply.
Phil

[stubbsonic]

Some synths are just one-sound-at-a-time creatures. You send it MIDI notes into its MIDI IN and it plays those notes on that one sound (assuming you are sending it MIDI notes on a particular channel, and it is set to receive on that channel). Most of the time, those kinds of synths have a MIDI receive setting on them, usually in the global or system settings, sometimes there's a settings page called MIDI. If you set the synth to receive on a particular channel, then it will ignore notes on other channels. If you set the synth to OMNI mode, it'll play all the notes that come in, no matter what channel.

MANY keyboard synths nowadays are MultiTimbral, which like I said means that they behave like a stack of up to 16 synths. You put drums on channel 10, you put piano on channel 1, you put a bass sound on channel 3, etc. etc. This is kind of important, as when you have your recording, you may be selecting sounds for each MIDI channel of that keyboard.

One more thing that can be a little messy. If you are recording percussion sounds (and will be recording MIDI using different sounds than when you mix) you might want to strategize how you will deal with drum sounds not playing back correctly. I often keep percussion parts separated out on separate tracks. When I open a kick track in the MIDI graphic editor and it is not playing a kick sound, I double click on the keyboard to select ALL the notes of that pitch, then I drag up or down until it is on a kick sound. (If monitor is on, it should play those notes as it goes).

Also, keep track of where the panic button is in DP. If you end up with stuck notes, you'll want to be able to shut them off without any stress.

[stubbsonic]

Panic Button = All Notes Off.

[MIDI Life Crisis]

Panic Button = All Notes Off.

I also like to use a sledgehammer.

[Phil O]

Thanks. That helps me get started.

Phil

[HCMarkus]

Yes, I always find a sledgehammer gets things rolling.

Bottom line on the MIDI Phil, is just don't record any program changes or assign a program number from within DP. On playback, the MIDI track you play enable will play back to whatever MIDI device/channel it is assigned to and whatever patch you select for that device/channel from the device itself.

I will suggest that the performer monitor sounds that are at least similar in envelope and dynamic response to those intended for final mix, as performance technique tends to be greatly influenced by the sound heard while recording.

[Phil O]

...just don't record any program changes...

Thanks. I just discovered that the input filter has a checkbox for patch/song changes (I've never had to use that window before). I'm assuming that's where I'll accomplish this. Yes?

On playback, the MIDI track you play enable will play back to whatever MIDI device/channel it is assigned to and whatever patch you select for that device/channel from the device itself.

That's the ticket. That's the most important part of this project.

I know these are VERY basic questions, but I use DP for audio only 99.9 percent of the time. To me MIDI is just a four letter word. (No offense intended MLC ) Thanks for all the responses, guys.

Phil

[monkey man]

Bang-on with the Input Filter thing, Phil.

While you're there, you might as well disable poly and channel (mono) aftertouch, MIDI CC controllers (volume, expression, portamento and so on) and sysex messages if you can find the check boxes easily; these flavours of MIDI data are all capable of interfering with your chosen sound when you'd least expect it. The CC (continuous controller) data could alter filter (HPF, BPF, LPF) settings, and the sysex data could dump a completely unrelated patch into the (patch) slot being played. Similar mayhem is possible from the other types I mentioned. Chances are you'll not even send this stuff to DP in the first place, but who knows, an accidental nudge of a knob or a flurry of frantic button pushing on the synth could result in spurious, un-needed data's being sent to DP and winding up on the track.

All you want, by the sound of it, is note-on and note-off data, along with the sustain pedal. Anything else might just as well be filtered, so get ticking!

[stubbsonic]

Yea. The only time a program change would be recorded is if you happened to change patches during recording. But yea, filtering them is a safe bet.

If you ever get MIDI tracks from a client, just open the event list for each track. You can see them in there and delete them if necessary.

[stubbsonic]

One other thing. As you record the MIDI tracks, you will be dealing with Velocity-Response. This is both how hard you strike the keys, AND how the synth responds to this. If there is a difference in how track sounds, dynamically, you can use DP's velocity adjustments. I use this one pretty often. It lets you kind of remap the velocity of the track to tweak the dynamics.

Region/Change Velocity.../Compress/Expand

Compress/Expand is one of the modes in that velocity editor.

[monkey man]

Even easier for Phil might be to instantiate the Velocity / Dynamics (or whatever it's called) plug-in on the MIDI channel in the mixer.

That way he'll have visual feedback and hands-on control of the dynamics more akin to what he'd be used to, i.e., using a compressor, and there'll be no need to even look at velocity data.

[Phil O]

Hey, thanks for all the feedback. This is very helpful.

Phil

[Shooshie]

Even easier for Phil might be to instantiate the Change Velocity plug-in on the MIDI channel in the mixer.

That's what I'd do if I were in Phil's shoes. The plugin will help you learn what the parameters do, and you can make it permanent later, once you've figured out how you want it.

Use the compression tool to alter the velocity same as you would compress audio: ratio, threshold, gain. It has a graph to illustrate what you're doing, which I always find helpful in compression plugins, even if they aren't necessary.

Velocity isn't volume, however. You have to remember that in MIDI, Velocity is just the first command to which an instrument responds for any given note. How the instrument is programmed to respond can vary widely. But if we're talking about a MIDI piano, guitar, or other such straightforward instrument, Velocity is merely the initial attack level, and you can control or compress it without any thought to what else it might do.

I would not turn off MIDI Continuous Data Controllers in the Input Filter or View Filter. Technically, they're called Control Change data, abbreviated CC, but I find it more intuitive to call them Continuous Controllers, also CC. Controllers can be essential to playing many sounds, and should be left in place. For those sounds, Velocity is still important, even though the instrument follows controllers for dynamics. Here's a brief description of the relationship of Velocity and Controllers to sampled sounds:

An instrument playing samples responds to velocity by choosing the dynamic "layer" of samples appropriate for each note at the loudness range at which you play it. Samples are originally recorded at anywhere from 2 to 20 layers (or any number). Each layer is recorded a little louder or softer than the next, and represents a specific dynamic level of playing throughout the range of the instrument. A 7-layer sample set has seven complete sets of samples, each set including every note on the instrument, recorded at a particular volume level. (seven layers would be: ppp, pp, p, mf, f, ff, fff) When you play the samples, 128 levels of Velocity are divided into 7 ranges according to our perception of those dynamics. When you strike the note, the range you hit determines the dynamic layer it picks to play back. (Example: velocity of 64, which would be right in the middle, would be mf, the middle layer — 4th layer in this case.)

Velocity tells it which layer to use at first strike. Controllers, then, tell it how loud to play that layer. As the note changes loudness, which is continuous in expressive playing, the CC data are pouring in, constantly telling it how loud to play. In a really good sample set, a crescendo with Expression (CC#11) or Breath Control (CC#2) will tell the instrument to crossfade between layers as it goes from mf to ff, so that you don't end up with a mf layer being played at full volume. Loudness is not just force; it's about timbre, and only by recording an instrument playing loud will you get a sample for what "loud" sounds like on that instrument. So, in playback, "loud" has to come from the loud samples layers, "Soft" has to come from the soft layers, and so on. Continuous data keeps those layers playing expressively and fading imperceptibly between themselves.

I explain all this, because so many sampled VIs depend on it nowadays. 10 years ago, most of them did not do the crossfade as you changed dynamics, and they sounded pretty awful. It's only complicated if you never try it. Once you see what it's doing, you go "oh, yeah! That's the only way it could really work!" This is still pretty basic stuff.

So, leave the MIDI Continuous Controllers in there.

Shooshie