VSTI world hate MIDI GUITAR?

[michu]

regarding S&H, when releasing finger from fretboard you most likely generate some noise that will fool a tracking algo for a moment, then it's frequency falls to 0 (infinite time before 0 crossings) and level dependant gate/env followers has some integration time for release, so it is very probable that sustaining tone will be off...
EDIT - an afterthought, of course it would be possible to delay a bit pitch signal to be held so pitch used for release is from before the gate shuts down and crossfade instead of instant switch,
heh, have to play a bit with that one
/EDIT

hehe, syncing oscs to input is even more responsive than pitch2cv as there is no integration of pitch information that is required by such algorithms. also, you can pull simple chords thru it with reasonable result
i was thinking about trying combination of pitch2cv so the oscs ain't fixed pitch and hardsync, always have something else too do

[MaxJoy]

The delay trick should work, good idea . With a tempo-synced delay you should even have a chance to re-use thrills etc. in a useful manner. That would end up in a delay effect that's synchronized to note release. Should be easily controllable. Sounds very promising to my ears .

[michu]

MaxJoy, tempo syncing makes not much sense, as with too much delay and a short note you could end up with using a silence before you played anything

[MaxJoy]

It's possible because you know the note length ... simply grap the good part of the delay buffer.

Ok, an ordinary delay can't do this, but David Haupt has a SampleSnatcher that comes to my mind as candidate

[MaxJoy]

The alternative is to use a normal delay with a very short buffer size. This way you always re-use the last part, this is much simpler and you will have less trouble with different velocities.

[kevink]

Obviously, some people really like that. So are you committing to do this, given that it's only a few hours work? When will we see a Poly-AnaFX demo? Will you provide a VST MIDI output? As you note, that's the "killer app" for pitch-to-digital. Jeez, I might even *buy* the thing if it could do that.

It's certainly a few hours work, but when everything else is done. So I'd say sometime after ver 1.0. Which is realistically still a few month's off.

You talk about *buy*ing software as if it's some rare exception?

Not at all. I've downloaded a fair number of free/donationware plugs, but the only ones that are really "bread and butter" for me are Ziokiller's stuff, and I've sent him donations. Otherwise I use mostly Sonar/Audition/NI/AAS/Virsyn/Arturia stuff, all paid for and registered.

It's just that I had written off buying, or even demoing, PolyAna based on the attitude you display toward your customers, as was exhibited explicitly in the thread discussing the PolyAna GUI and implicitly in this thread. Someone who can't take constructive criticism, and who puts his ego ahead of the needs of his user base, is generally unlikely to produce a quality product or provide quality support. Besides, I've already got a pretty good collection of VA plugs and don't see much need for yet another one, particularly if it's burdened with such an impractical GUI. If it actually did something that nothing else in my collection does, such as doing real-time pitch-to-MIDI as an effect, then I might consider it.

[AdmiralQuality]

It's just that I had written off buying, or even demoing, PolyAna based on the attitude you display toward your customers,

WHAT customers?

as was exhibited explicitly in the thread discussing the PolyAna GUI and implicitly in this thread. Someone who can't take constructive criticism, and who puts his ego

Poly-Ana is not designed by committee. Did somebody make you think my company's business is a democracy that you're a citizen of? If you don't appreciate it, FINE, it must not be for you. But jeeze... if I felt the need to jump on every product I'm not interested in with negative comments I'd never stop typing. (And I only commened on FS's stuff in response to guitarzan's statements that all other types of MIDI-guitar are "shit")

ahead of the needs of his user base, is generally unlikely to produce a quality product or provide quality support. Besides, I've already got a pretty good collection of VA plugs and don't see much need for yet another one, particularly if it's burdened with such an impractical GUI. If it actually did something that nothing else in my collection does, such as doing real-time pitch-to-MIDI as an effect, then I might consider it.

Well I was just discussing adding that for y'all. Wasn't I?

[nix808]

Do you think realtime pitch to midi conversion is technically possible, Admiral Quality? This would be an awesome device if so, for players of instruments other than piano or guitar. I think that no matter how technically gifted you are, you wouldn't get any better performance out of a GI-10 than Sascha, if you didn't modify it's innards. I've experienced different performances from different midi units now, and the GI-20 simply tracks better for midi than the GR-20. Thanks for your help.
Happy music making, Nix

[AdmiralQuality]

Do you think realtime pitch to midi conversion is technically possible, Admiral Quality? This would be an awesome device if so, for players of instruments other than piano or guitar. I think that no matter how technically gifted you are, you wouldn't get any better performance out of a GI-10 than Sascha, if you didn't modify it's innards. I've experienced different performances from different midi units now, and the GI-20 simply tracks better for midi than the GR-20.
Happy music making, Nix

I've got my heart set on a GI-20. But again, can't afford anything right now. Broke and I'm being forced to move sometime in the next few months... our building was sold.

And to answer your question; Do I think realtime pitch to midi conversion is technically possible? Nothing is realtime. Like I was saying last night, even the sound from real instruments takes time to get to our ears... especially if you play through amps vs. acoustic guitar or piano that's right next to your ears.

But do I believe FAST pitch to MIDI conversion can happen? Sure. Like I keep saying, there's no reason FS's synths couldn't output MIDI of the closest-to note followed by a thick stream of pitch bends to correct to the exact pitch as detected. Really, I could code this in half a day and maybe I will for you guys, just to make some "friends" back from all the surly guitar players I seem to have pissed off here. I'd have it also send Env Follow controller messages which you could route to some other parameter, like filter cutoff or amp gain.

NOW... aside from MIDI's bandwidth problems (which actually don't exist internally on your computer, only when it goes out on an actual MIDI interface) there's one other problem with VSTi and how they treat live MIDI. The issue is, that MIDI events can only be received between the sample chunks that you're processing. So the larger your buffer, the more error in real time MIDI performance you'll have. Even between VSTis running in the same host.

I have a 3 ms buffer. So it's not too bad. But MIDI can potentially send a new message every.... (does some math again)... about 0.9 ms! And 0.6 ms if it's the same type of message as the last sent message (called running status.) And imaginary in-host MIDI messages have no real bandwidth limit, so messages could POTENTIALLY come every sample. But the VST instrument or effect can't receive any MIDI or automation events during the processing of the buffer. Only between. So if you have larger buffers, you'll tend to increase "zippering".

Now, in non-real time, the MIDI messages have sample position offsets which DO allow them to become sample accurate. So you could actually get more accurate output during rendering than live. But whenever played live, these offsets are always zero because every message recieved is already inherently "late". (To specify the sample position, the host needs to know in advance that the note is coming. So only works for pre-recorded sequencer playback.)

Anyway, it's not a BIG deal, and it can be mitigated to a great deal by a little smoothing of the controller streams done by the VST instrument itself, though not many actually do that. But do keep your buffer as small as you possibly can. It's also the cause of latency, so most of you already optimize this as much as you can.

So FS's stuff doesn't have this limitation, his CV can potentially change with every sample. There's a big plus, and something that IS superior to MIDI. But the big minus is you're stuck with only the sounds of those instruments and can't control others. But like I said, if you like them, and only them, then you're all set.

Now, I really don't think the naive method of pitch detection is best, but it's certainly easy to implement and would be a worthy experiment. So yeah, if I can find some time in the next couple weeks I might toss off a little free one for yuz. Of course, half the people here won't be able to use it as they're boycotting my products.

[nix808]

So you think it could be done workably? I think you could be the prickliest of characters, and still sell this device. It's what I was looking for before I started looking at midi guitar. Do you have any ideas as to how to implement a pitch to CV or pitch to midi in Synth Maker? I'd ask Fretted Synth, but I don't think he'll coming back to this thread(understandably).
Cheers, Nix

[AdmiralQuality]

Actually I'd do that one for free. I wouldn't consider a "naive method" pitch tracker to be worth anything.

As for Synth Maker, never used it. Though it looks rather nice. It seems to have scriptability too so you'd probably just write a script to do zero crossing timings.

Or if you know C++, just write your own plug-in module. But then if you can do that, you don't need SM or SE.

[nix808]

I am about as fluent in C++ as I am in Innuit (not that they're an unimportant people). Is there anything wrong with this naive method of pitch tracking? So if you can code a naive pitch tracker, you can then extract midi pitch from the zero crossings? I don't really know anything about DSP, but I love computer music for it's possibilities, and would like to see this tool come to be. I think it would certainly be marketable, if there was any accuracy to it's tracking.
Cheers, Nix

[AdmiralQuality]

Well it's just that timing zero crossings isn't the best method because waveforms aren't always symmetrical, and they don't always have only one zero crossing per cycle of the fundamental frequency. Depends on the strength and phase of the harmonics.

This is why FS's stuff recommends you play on the neck pickup. You'll get a "rounder" waveform there.

A slight improvement is to low pass filter the waveform first. But the optimal amount is different for different pitches... and if you knew the pitch you'd be done.

If I do do this (and I probably will) I'll spend an hour or two seeing if I can't come up with a better method. Just was thinking of some fairly simple things that would probably work.

There are certainly other pitch to MIDI plugs out there already. Probably even free ones. But I don't know if anyone's sending envelope as an additional CC which would be very useful for guitarists.

[nix808]

Lateral Solutions Guitar Synth is one free one(sadly). It's not bad, but it plays multiple notes a little, or there were some errors as I recall. This may be due to asymetrical waveforms you mentioned. So the naive method to extract pitch data from an incoming wave has it's shortcomings, I gather that's still how you plan to do it? I'm still interested to try and see if I can create an audio triggered synth in SM. Also at any ideas you have had to implement this, though I wouldn't blame you if you kept them to yourself.
Cheers so.

[michu]

AQ, once, for fun i tried fitting the gk2a pickup upside down and was surprised how steep was filtering for each string, both lowpass and highpass (this was with ancient gr9 unit).