Polyphonic Guitar to MIDI VST/AU "MIDI Guitar"- BETA TEST

[FrettedSynth]

Testing v0.4.2 and again I have to say WOW

Thank you for the octave switches they make it much more easy to use quickly. All setups seem to save well now, v0.4.1 had broke this.
I really like the per note sensitivity settings each with its own level meter.
The fixed velocity will also come in handy, Thanks

@metalifuxx great job glad to see you still at it
If you ever find yourself with a drummer around but need a bass player to fill in so you can jam, Let me know I would like to jam with some of the KVR people I have met over the years. I am not far north of Detroit.

Peace
Joe

[Eleventh]

Thanks for the updates. Still one problem with preset saving though: whenever I load the plugin, it always loads the test piano as the instrument, even though I saved the preset with no instrument selected.

Cheers.

[insaneacyde]

oh awesome, just got the early adoption e-mail
very excited about this product, gimme a bit to test it but just reading the positive feedback here has got me uber excited!

[spinedoc]

OK, just got this working in Sonar - previously could not get Sonar to recognise its MIDI output, as Enable MIDI Output didn't do so, even when MIDI Guitar was had "Configure as a synth" checked under VST properties.

Sadly, the only way I could get Sonar to play ball was a registry hack, albeit a minor one.

The key is how Sonar recognises it on initial scan, so dig into
HKEY_CURRENT_USER\Software\Cakewalk Music Software\SONAR Producer\Cakewalk VST X64\Inventory\
and find the MIDI Guitar entry, then edit the generateEvents key to have the value 1 instead of 0. (Usual caveats apply re

Whether this is a Sonar shortcoming, or whether there's something in MIDI Guitar that isn't set up so that SONAR is getting this information, I don't know. The net result, though, is that Sonar now shows a track with MIDI guitar (even just playing an audio file, rather than dealing with live input) as a midi source in soft synth tracks and midi tracks.

Case solved! Hopefully.

Nice job on this hack. I've seen something on the Sonar boards about this but haven't had time to try yet until now and it works great. For those users on 32 bit just do a registry search for the generate events key and it'll come up. It is not under the x64 obviously.

I am wondering what type of Gen. compression settings people are having the best luck with when placed in front of midi guitar. I understand it based upon a signal input on the guitar it will be different but generally What attack and release settings have been best for you. So for the fast Attack seems to be generating better response from me but I'm just beginning to fiddle around with this as a pre-stage in front of midi guitar and am going to start messing with exciters next.

[JamOrigin]

I am wondering what type of Gen. compression settings people are having the best luck with when placed in front of midi guitar. I understand it based upon a signal input on the guitar it will be different but generally What attack and release settings have been best for you. So for the fast Attack seems to be generating better response from me but I'm just beginning to fiddle around with this as a pre-stage in front of midi guitar and am going to start messing with exciters next.

Thanks for looking into this.

I think its a triple-edged sword. The loudness gain is good, but the lowered dynamic range is bad, and of course latency could be bad. The latter you may not have control over, but the former two you somehow have to balance out. A fast attack would generally work best.

MIDI Guitar does already a special kind of compression, but we have not exposed its parameters to let you control it, so I'd think that when you (and others) find compression can help, its because your can tweak those settings to suit your particular sound.

Im very curious about your findings

[metalifuxx]

I am wondering what type of Gen. compression settings people are having the best luck with when placed in front of midi guitar. I understand it based upon a signal input on the guitar it will be different but generally What attack and release settings have been best for you. So for the fast Attack seems to be generating better response from me but I'm just beginning to fiddle around with this as a pre-stage in front of midi guitar and am going to start messing with exciters next.

Thanks for looking into this.

I think its a triple-edged sword. The loudness gain is good, but the lowered dynamic range is bad, and of course latency could be bad. The latter you may not have control over, but the former two you somehow have to balance out. A fast attack would generally work best.

MIDI Guitar does already a special kind of compression, but we have not exposed its parameters to let you control it, so I'd think that when you (and others) find compression can help, its because your can tweak those settings to suit your particular sound.

Im very curious about your findings

I tried this originally in version .2 when I first tried the plugin, using a couple different plugin compressors, one a guitar specific compressor being Distorque's Orange Squeezer and also the free Klanghelm DC1A, on various settings. I found just getting a good strong almost clipping signal with my interface's (M-Audio MobilePre) preamps got the best response/reaction/control and sustain the plugin was "looking" for.
Maybe some sort of signal normalizer or gain booster works better than compression. I'd think in most cases you would want natural dynamic response from your guitar so it properly translates the sustain and velocity values sent to the vsti/instrument. I guess depending on the final vsti sound dynamic/character/tone you are looking to get, would determine the use of a certain pre-signal compression? Also depends if you are playing single note monophonic riffs or polyphonic chord and arpeggiated chord picking. A full strummed agressive played chord will give more "power" than single picked notes usually, and I have found that also affects the plugins note/velocity/sustain translations when my preamps are set at different levels.

[JamOrigin]

Ok so here is the previous idea in a whole song context with more layering.

Bassline and vox/choir all played through Midi Guitar. No keyboard/midi controller, programming, and no midi loops were used whatsoever. And I certainly can't sing like that either

The Natty Daddy Choir (an ode to the cheap malt liquor and jamming with good friends):D
http://dl.dropbox.com/u/6185854/kvr%20songs/ND.mp3

Thanks for posting this, metalifuxx.
This makes me wonder whether it would be nice to have a 4-track recorder/looper (both audio and MIDI) inside MIDI Guitar at some point to put down ideas quickly?

I think you are right regarding compression. In most cases you'll want to stay with a natural dynamic guitar response, but I could be wrong.

Eleventh, thanks for posting the problem with the Test Piano - we sorted it out now for next update.

[gregh1]

The plugin could have an envelope follower on the (pre-compressed) input, then compress to feed into the MIDI conversion, then apply the envelope from the original signal to the output MIDI events (note-by-note, not as CC 7 values). This way, the dynamics of the original signal are preserved. Maybe the plugin already works this way.
If the user wants less dynamic variation in the MIDI output, then an additional compressor before the plugin could be used.

[AdmiralQuality]

The plugin could have an envelope follower on the (pre-compressed) input, then compress to feed into the MIDI conversion, then apply the envelope from the original signal to the output MIDI events (note-by-note, not as CC 7 values). This way, the dynamics of the original signal are preserved. Maybe the plugin already works this way.
If the user wants less dynamic variation in the MIDI output, then an additional compressor before the plugin could be used.

That will probably make sense once they implement multichannel (per string) output.

The only per-note continuous controller MIDI offers on a single channel is polyphonic-pressure/aftertouch. Yes, maybe they could implement an envelope follower output right now on the single channel output as poly-pressure. Aftertouch/pressure is most often used to modulate filters and/or volume, so that makes the most sense.

And once multi-channel output is supported, you'd probably want to provide a feature to switch that to channel aftertouch, as not every instrument supports polyphonic aftertouch. And you'd probably want to be able to assign it to any CC# as well, as once it's providing one MIDI channel per note/string you're no longer limited by control change messages applying to the entire channel. Each string gets its own channel, so any of the CC numbers become polyphonic.

[JamOrigin]

gregh1's point is just that although we use compression internally in the MIDI conversion we can use the original signal for determining velocity etc...
This is basically what happens already, abeit compression of a proprietary kind. We do, however, have to look more into determining velocity during those few millisecs of the onset.

AdmiralQuality, AfterTouch could easily be added too. Good point with the multichannel output (which we also need for bends and micro-tunings etc).

[JamOrigin]

Now, we are talking of this, I'd like to toss in a related topic that some of you should find interesting. It's a topic that AdmiralQuality was kind to discuss with me a coupe of days ago, in private.

Obviously MIDI Guitar as a controller, has some interesting properties that other MIDI controllers (read: keyboards) do not.

MIDI Guitar will send MIDI messages as a keyboard, but in additiion it has the guitar audio output that we haven't talked about in this thread. In fact it has 6 audio outputs - carefully dissected out of the polyphonic signal. Those 6 outputs will contain the quasi-periodic signals from each string.

If quasi-periodic waves are not interesting as such, we can reduce it to loudness information with a enveloper follower, or perhaps loudness/"brightness" tuples or even track the dynamics of each harmonic produced by each tone in a multi-dimentional data structure.

Regardsless of whatever representation of this analog audio signal it has some profound musical properties that a midi keyboard does not have. Let me just try to give the simples example i can think of:

Apply the guitar-string envelope as envelope for each note in the synth. Even in this simplistic case the guitar (as a controller) offers something that the keyboard cannot. Presumably a complex, yet very controllable, natural real world physical envelope, that you can actually feel in the guitar body and in your fingers when you play it.

I'm curious whether you guys could come up with interesting ways to use such "analog/real world" data to drive synth dynamics?

[Eleventh]

As far as I know, the Axon MIDI converter is doing something like that, as it can detect where you pick the string - is it closer to the bridge or to the neck.

If you can track the brightness accurately, maybe one could then use the tone pot as a mod wheel?

What would also be cool is if the system would be able to distinquish palm muted notes from normal ones.

Btw, is it really possible to detect on which string you're playing the note by only analysing the monophonic output of the guitar?

[jas]

I'm very interested in this thread and have followed it for a few months. I play many acoustic plucked world instruments from many different eras and in varied tunings. Some of the instruments have tranducers mounted on them, but mostly I enjoy using the interface with microphone positioning and acoustics for varying degrees of interpretation. For the past two years, I've been triggering software synths (mostly pads) using Widisoft for real time recognition.
I realize the limitations of this as a controller, but have had good success by adjusting several controls in polyphonic mode using an Euclidian algorhythm. Control of the degree of accuracy is regulated by many factors - most importantly: equalization presets based on the range and harmonics of the instrument, as well of control of the degree of polyphony and noise gating employed. There are many other variables (such as minimal note decay) involved.
But I like seeing the development of your software, and honestly find that you are evolving around similar findings. What I mean by that, is that it is the nature of the instrument - not just the triggering mechanism - that allows for interesting compositions to come from synths as they follow the controller (plucked instrument) lead.
So, a big yes for audio to midi!

[dumant]

Hi,

I have tested your application on my Mac, and it really looks promising.

But I am unable to run it as a plugin (either VST or AU) in Reaper: The "I accept" button for the licence cannot be clicked. I also tried with Mu-Lab (crash), and Multitrack studio (same behaviour as Reaper).

I actually managed to use it with Reaper by using ReaRoute/Soundflower to send the audio to MidiGuitar, and IAC to get the midi back. But it's a bit cumbersome to set up, and doesn't help keeping latency low... If you can do something, that would be great !

Best,

Bruno


PS: my configuration: Macbook Pro Core Duo (MacBookPro1,1 2006), Mac OS X 10.6.8, latest Reaper (4.31)

[jas]

Can Midi Guitar be run as a standalone application that outputs midi to (for example) the IAC Driver on a Mac? I ask because in that way it could supply midi to multiple hosts or via Rewire or other standalone applications either alone or in tandem.