MIDI-only version of MG3?

BETA TESTING
1

First up, bravo- the pitch to midi in v3 is vastly better than v2. The MPE is also an awesome addition. All in all - an awesome achievement. I assume its done with machine learning - I would be very curious to learn about your training data, but that’s probably quite secret squirrel?

To anyone that hasn’t used MG3 with something that understands MPE - jump in fast - it is next level amazing with MPE. I particularly love it with Moog Model D and any Audio Modelling stuff.

However, I am wondering if you are ever going to produce a MIDI-only version of MIDIguitar, that focuses only on player settings, midi routings etc etc. I only really use MG2 in this way - I turn off everything that makes a sound - and I am doing the same in the MG3 testing. From where I sit, the real value of MG3 (to me at least) lies in its ability to make usable MIDI from a guitar signal. I am pretty uninterested in everything else. Anyone else feel this way? I believe that the pitch-to-midi is your main product - I would love to see you focus on just that.

And now to venture off into idle speculation. The Fishman TriplePlay software (ghastly though it is) has a “fingerstyle” setting - the latency seems to go up a bit when this setting is used - but it tracks fingerstyle playing a bit better. Rather than have the one algorithm to do everything in MG3, is it possible to have several, and for them to be loaded/unloaded as required? Different pickups even (humbucker/single coil/bridge/neck/in phase/out of phase). Maybe one for the distant future.

OK, to finish. I LOVE LOVE LOVE the pitch-to-MIDI. I think it really kills everything else now - I have already sold all my 13-pin stuff as a result of MG2, and now I am thinking of getting rid of my Fishman stuff too. Go you!

5 Likes
2

I totally agree, for me its also the main use for MG, most musicians probably have some setup they use anyway. However, what would a MIDI-Only version do for you? Should it be less expensive? Otherwise, what keeps you from just making a “MIDI Only” patch, just like I do with MG2 and MG3 now? Sorry if I’m misunderstanding, just interested.

1 Like
3

Exactly. It’s easy to overlook, and my hope is that MPE is a guitar-thing now.

The main focus is of course guitar-to-MIDI and the main idea with the MG3 application was to make midi-routing both easier and more versatile and consistent. Which is also needed to deal with the additional complexity of MPE. So I think it already is focused on midi routing, but of course currently maybe 70% complete…if you have any idea of improvements or stripping away something, please let us know!

As @normen said, all slots can load MIDI or audio modules, you can create your patches without any audio modules. But I see the Mixer and master chain is superfluous in your case. So, let’s make a way to create patches without this.

Returning, to your quote above… we have a challenge to communicate to new users, all the opportunities MG opens up. As you mentioned, people won’t otherwise notice for example MPE. Generally, the way we can show the possibilities, is by putting a few synths on board and make some factory patches with synths/audio.

There is also an argument about latency and easy setup: with the standalone app directly hosting audio plugins, less things can go wrong and there is a smaller latency overhead. This way, the standalone app is a live performance host, more lightweight than a DAW.

Regardless of whether you use MG as a pure guitar–to-midi plugin in a DAW, or as a standalone performance host, there is no compromise. It’s not like anything would improve on either side, if we cut out the other side. Hope this makes sense.

To your last point. It may be possible to adjust tracking to accommodate speciffic playing styles or even fine-tune the tracking model to your particular style and guitar sound.

2 Likes
4

I agree completely. I wish they would focus their work on the algorithms that analyse and translate guitar audio to MIDI accurately. Latency being the Whole of the Law.

With mg3 they seem to be trying to turn it into a ‘do everything’ utility.
I don’t need that. I have a good DAW and plenty of VST instruments already available.

1 Like
5

I agree also.
It does seem like the project has gotten hung up on trying to include too many internal features, and provide the app on on too many platforms. Using Gig Performer, I have no need for any internal features other than strict, low latency audio-to-MIDI translation. That is the whole point of the app—at least, it was.

Since I use drop tunings, MG3 has yet to be of any use to me in that regard. The bread and butter of it isn’t as useful as MG2 is to me yet, and considering how long this project has been going on for, it’s very puzzling as to what the actual goal here is.

My advice and comments to the Developers: You cannot be everything to everyone, all the time, everywhere. You just don’t have the manpower or resources to do so in a timely way, and that should be abundantly clear by now. It’s not a crime or a harsh criticism, just a simple fact. Pick and choose the areas for improvement over MG2—namely MPE inclusion and improved tracking, and make a public release available for Mac and PC. You can always add features in future releases and updates, as well as inclusion for other platforms—if there is even a reasonable demand for such a thing.

I have enjoyed using and supporting the MG products for years now and will continue to use MG2 regularly until such a time as MG3 is shown to be more functional at the things it should do best.

6

to present a contrary view: everytime jamo goes fishing he lands a pretty big one.

i’m happy to relax a bit and let him experiment and explore, in fact i strongly prefer this.

things like hex or net or the (possibly forthcoming as hinted and surely to be truly majestical) autotune are like gifts from heaven.

i’m in no hurry.

2 Likes
7

I understand the sentiment here and clearly it’s something that comes up now and then. Using MG just for getting midi into a DAW is surely one of the most important use cases. But there are also people who use MG on stage, and people who are guitarists-first, who never used midi and never had any reason to install a synth. There are users for whom latency is everything, and ironically they are better of using MG3 audio features (ie. load an external synth into MG3) and add a Midi Output to send it on to the DAW for recording midi. There are users who will never tolerate the discomfort of midi, who use the Deep Expressor and add synths just for ambience. Possibly the majority of users need help getting started with MG and just to get sound out of it. And there are people who can be in all these categories but like to be able to switch among them with a Patch change.

But all that said, there is certainly a good case for a midi-only plugin. As a first thing, I’ll make sure, the MG3 VST/AU will load the MINI patch at startup, in all DAWs. This way you will never see any audio features, unless you change its patch. Let’s see if that cuts it.

Just be careful of assumptions:

If such a plugin could reduce latency by 1ms, make it easier to setup or save a few CPU cycles - we would have had it 10 years ago. I’ve been in this long game of improving tracking for almost two decades now, so rest assured, i’m not leaving 1ms of latency or 1 cent of pitch resolution on the table. A midi-only thing doesn’t magically become “better” in any technical way.

Further, a midi-only plugin will have to be an audio-fx! Thats because it operates on guitar audio and the only way to get that reliably from DAWs is as an audio effect.

Further, with a midi-only plugin, you may be surprised that you actually would want serveral midi modules (such as Legato, Chromatic, Dynamics, Modulators)… and be able to add CCs to the Midi Output, because you can’t program that in your DAW. Since MPE its hard to concieve of a small beautiful black box, without making compromises.

Lastly, complete disclosure: the trends in machine learning and “AI” are the opposite of the sentiment here: in the future we will have better things (less latency, more responsiveness, more expressions, etc) because of end-to-end (ie. audio-to-audio) methods.

9 Likes
8

@JamO, I can only fully support your point of view!

At the end of the day, it’s all about giving musicians (and listeners) the feeling that it sounds organic and musical. A psychoacoustic approach that artfully combines parts of audio, audio modeling and midi sounds could contribute to this.

9

I can see that for live work there is a good case for a standalone utility that can trigger third party VST instruments for sound generation. You don’t want the added complication of a DAW in that case.

For recording I do like to have the MIDI track explicitly available, though: I’m used to doing quite a lot of rather detailed MIDI editing. In fact I’m becoming more of a composer than a player these days, I guess.

So I don’t really know what to say about splitting out the MIDI conversion as a separate product. As you say, it’s not going to improve the MIDI conversion as such. The only thing I might mention is that we impoverished musicians are not made of money, and might be more willing to pay for a more targeted product that is somewhat cheaper?

10

I’m not saying that MG3 or any simplified version of it has to be necessarily better at latency. MG2 has good, useable latency for live performance already. Accurate tracking of notes is probably the more important feature where strict audio-to MIDI is concerned.

I still stand by my earlier statement: You cant be everything, to everyone, all the time, everywhere.
I appreciate your efforts and interests in covering all the areas, but there still have to be practical goals and priorities within a reasonable time frame for the product to have success. As a fellow ‘mad scientist’ when it comes to music creation and engineering, I understand the unending need to tinker and improve upon a creation, but I also know that when things remain a work- in-progress for too long, they reach a point of diminishing returns where a disproportionate amount of time is spent on a small fraction of improvement. It’s an easy thing to lose sight of.

As I said before, I have enjoyed using MG2 and look forward to whatever you come up with in the future. I’m simply throwing in my perspective here as a consumer that the goal of a ‘finished MG3’ has become more and more unclear through the MG3 beta cycle as things continue to get added.