A Midi-Editor for Guitar and Other Stringed Instruments

[Christian_Schormann]

Creating guitar parts with the piano-roll editor is painful. Controlling the proper fretboard positions for notes usually requires either key switches with offset timing or the use of multiple channels, either of which is uncomfortable to do, and time consuming and error prone. Chord strumming parts are even more painful. The score editor is not really an answer - it is still very slow for my purposes.

I propose the addition of a specialized editor for guitar parts that makes it easier and faster to create good, realistic guitar parts in Cubase. I wrote a working prototype of the editor that demonstrates the concepts I am looking for - screenshot below, and then a more detailed explanation.

The editor is similar to the piano-roll editor, but shows strings instead of keyboard lines. Bars on a string indicate a note. A number on the bar, just like in a guitar tablature, controls the fret position on the string. When dragging out the note, vertical drag lets you control the fretboard position as you go. The orange bar below the note controls velocity. Other expression parameters can be set with an RMB shortcut menu (design in progress) - you see some visuals below for the legato curve or the hash patterns that indicate mute playback.

To make editing fast, all the usual cut copy paste functions, including option drag for fast copy, as well as rolling edits where notes touch, are available for very fast editing.

Of course, guitar parts often mix picking and strumming. Strum parts are not fast when created from individual notes. To make this easier, the prototype adds a chord library (which could easily be user-extensible, although the prototype does not implement that yet). Using the library, you can search for a chord and drag it to the timeline (or you can type in a sequence like “G 4/4 Dsus4 8/4” in a text box to insert a series of chords rapidly). For a chord on the timeline, you can quickly create a strum pattern drawing bars just like for notes, where you can control the type of strum you can do, down, up, slow down/up for arpeggios, muted strumming and so on. There will be a library of strum patterns soon, just haven’t gotten to it yet. The following example shows a chord in the editor, complete with a chord diagram (which determines what is played) and the strum pattern above in orange. Of course, instead of the letters for the strum type, there should be icons like up and down arrow etc. You can also separately control which strings the strum plays, as often guitar players will not play all strings of a given chord. Chords can also be replaced easily with drag and drop.

Here is a screenshot of the current chord library. The search bar needs a bit more design work:

The current prototype loads an AU plugin (MusicLab RealStrat) for interactive feedback when drawing and playback. It maps expression features to the available key switches etc. in RealStrat. In a real-world implementation, that mapping could be made configurable. I have started that work for test purposes, and I think it is feasible (at least for a few of the instruments I have), but admittedly there is more work to be done here. It might also be worth to propose a standard way, for example for Midi 2.0, to encode Guitar expressivity in midi, so that such mapping work becomes unnecessary. Finally, Steinberg could of course deliver a custom guitar instrument as part of Cubase

I’d also be super-happy with a plugin interface that would allow me to create an editor like this myself and run it inside of Cubase…

Any feedback on the design? I am more than happy to share more details or discuss.

Thanks a lot for reading.

[Christian_Schormann]

One additional comment: Some guitar plugins, like Ample Sound for example, of course have editors based on a fretboard design, although I believe that this proposal is a bit more extensive and productive.
But the core issue for me is that I don’t generally want part editors inside a VST. I want editors adequate for the instrument I write for in the DAW, so that all my arrangement activities happen in one place, in one timeline - the Cubase timeline.

That is also why I would very much hope that Cubase might introduce a plugin API one day that allows users to create their own custom editors for Midi-parts on the timeline. Special purpose editors rock - but at least in my opinion they need to operate on the parts on the timeline so that all arrangement, part copying cutting etc. continues to work in the context of the song, as opposed to a separate instrument window that is unaware of the song structure.

[rickpaul]

This is an interesting concept, and, as a non-guitar player who does all my guitar parts with virtual instruments (MusicLab, Acoustic Samples, NI Session Guitarist, UJAM, Chris Hein), anything that could make it easier to create expressive guitar parts could be welcome, though I am generally doing that initially by playing the parts on a keyboard and I rarely program anything in a MIDI editor, though I do end up extensively editing what I’ve initially played for these specific purposes, and it can indeed be painful.

One key thing I wonder about with your specific design suggestion is how intuitive it will be to someone (like me) who is not a guitar player. (Just to qualify the “not a guitar player” thing, I can play a little bit of guitar, but I haven’t had access to one in a decade and a half or so, and I could mostly just play some melodies and a few chords, but I could never really switch between chords well enough to even play simple songs reasonably, no less did I know all the expressive things that real guitar players would do.) I more or less understand tablature, though I could never read it as quickly as I read regular musical notation. It might be that I’d adjust to it well enough over time, though a key would still be how do I get from playing something (on a keyboard) to tweaking it in this model (e.g. maybe adjusting which strings, hammer-ons/pull-offs, bends, etc.).

@Christian_Schormann

Of course, guitar parts often mix picking and strumming. Strum parts are not fast when created from individual notes.

Yeah, definitely. Of course, I play both from a keyboard now, and this is part of my question here – how to get from the chords I play that would be strummed to realistic guitar string mapping. Some of the virtual instruments do that more or less automatically, and/or with keyswitch controls to modify the default choices. But you also get into the territory where, for example, MusicLab has a zillion controls, and that is part of what makes this stuff so tedious to work with when needing more expressiveness than what comes off playing the keyboard and any played (or overdubbed) controllers. For example, let’s say I’ve played a part that mixes some melodic playing with some strummed chords. Now I want to refine the part, does the chord diagram side of things allow me to select some notes that I’ve played, which I recognize as, say, a Dsus chord, and replace it with the notes from the relevant chord diagram so it gets properly voiced?

@Christian_Schormann

There will be a library of strum patterns soon, just haven’t gotten to it yet.

This is an area where I’ve found the virtual instrument pattern facilities to come up short, thus leading me to either use loop-based strum instruments (e.g. UJAM or the patterns side of Session Guitarist), or just play the chords and the up-down strokes (e.g. MusicLab), but both alternatives tend to come up short. The latter never quite feels totally realistic to me, and there is a lot of hunting and pecking to find the right patterns and the right times to switch patterns in the former. What I’d optimally like (whether or not it is a realistic wish list item at the moment) is that I play the rhythms I want, and they somehow get translated (be it with manual, but intuitive, editing or with some AI-type analysis) to authentic-sounding guitar parts.

@Christian_Schormann

The current prototype loads an AU plugin (MusicLab RealStrat) for interactive feedback when drawing and playback. It maps expression features to the available key switches etc. in RealStrat. In a real-world implementation, that mapping could be made configurable.

I think the configurability would be important, but, even better yet, would be if there were some arrangement to get key virtual instrument developers to do this sort of mapping for their instruments, or maybe if Steinberg were to provide such mappings (probably only feasible for a limited set of vendors and only if their multiple products had the same expression capabilities maps).

@Christian_Schormann

Finally, Steinberg could of course deliver a custom guitar instrument as part of Cubase

That’s one I’d probably “veto”. While it’s not a bad idea to make something available, realistically, there’s no way a single virtual instrument could satisfy all the needs any give user might have for virtual guitars, and thus the feature wouldn’t be very useful if it didn’t come out with support for key third party virtual instruments. (Of course, it should also support Steinberg’s own products like TGuitar.) This is the sad fact behind why users like me end up having a zillion different virtual guitar products (e.g. in my case all the MusicLab products, quite a few Session Guitarist products, all the UJAM VG series, and assorted others). Any new Steinberg custom guitar instrument would need to at least be in the same class as these dedicated products, but, even if it was, there would still be the need for the feature to work well with third-party instruments.

[Christian_Schormann]

Hi Rick, thanks for the long reply. I appreciate it.

This design, I will admit, is primarily about creative control and ease of experimentation - making use of the detailed creative control that instruments such as MusicLab or Ample Sound offer, without the pain and time investment this causes in a piano roll editor. But it is true that making use of this creative control requires at least some idea of how guitar fingering works.
That said, even with very basic guitar skills, I think the design makes it a lot easier to experiment with fingerings and expressions, which in turn should make it a lot faster to get to good-sounding guitar parts, when compared with the piano roll editor. I find the collection of extra key switches and/or timing adjustment required to get the key switches and notes in the right order extremely cumbersome, and very fragile when revising an initial part that worked. The editor encapsulates all that technical detail into the musical/guitar-ish concepts behind, which I think is much easier to understand and maintain.

You do point out one thing that my prototype does not do yet (thanks!!!), but which should actually be fun to experiment with: Taking a recording (just pure midi without any helpful hints) and finding a default fingering. I just recently found some algorithms that might be helpful in doing this automatically (e.g. GitHub - natecdr/tuttut: Converts a midi file to guitar tabs (or any fretted string instrument) · GitHub), but I haven’t experimented with that yet. I may try that if I find the time.

For chord patterns: I agree that with libraries, the hunting problem is always pretty daunting. I don’t have a good solution for that, but I do think that searching based on a rhythm you play should be quite feasible, which might make a library more useful.

Perhaps more importantly, by having a strum editor that focuses on the musical action - how a guitar player would strum, including expressive variants, I believe makes it much faster to either draw or record patterns and then edit them till they work, even without a library. And it invites experimentation - for me, easy experimentation is key to the creative process.

One thing I did not mention is conversions - the prototype already can convert chord strums (including patterns) into notes that can then be edited in detail. That is often much faster than drawing, and for me it’s often even faster than recording. It should be possible to do the inverse as well (taking a recording and creating a chord pattern), further speeding up the correction process of recordings. And for the instruments where key commands are used to control strumming when playing (as opposed to actually playing the notes), these should be easy enough to convert into strum patterns.

That of course does require a really good mapping editor

To your point regarding an instrument: I did not mean to suggest that Steinberg should have an exclusive instrument, but I do think it would be good to have a default instrument to go with the editor (which might introduce a default, “standard” mapping). Of course, you would still want to be able to create mappings for other instruments. I have a gazillion of these instruments too, in addition to my real guitars

[rickpaul]

@Christian_Schormann

This design, I will admit, is primarily about creative control and ease of experimentation - making use of the detailed creative control that instruments such as MusicLab or Ample Sound offer, without the pain and time investment this causes in a piano roll editor.

I’m certainly all for that. I might add that I’ve tried creating expression maps to help out on this front, but I’ve mostly failed miserably.

@Christian_Schormann

I think the design makes it a lot easier to experiment with fingerings and expressions, which in turn should make it a lot faster to get to good-sounding guitar parts, when compared with the piano roll editor.

And experimentation, with the notion of “I’ll know it when I hear it”, can indeed be helpful.

@Christian_Schormann

I find the collection of extra key switches and/or timing adjustment required to get the key switches and notes in the right order extremely cumbersome, and very fragile when revising an initial part that worked. The editor encapsulates all that technical detail into the musical/guitar-ish concepts behind, which I think is much easier to understand and maintain.

Yes, however, the key really is a combination of making sure the facility can map to the individual keyswitches and CCs used in key virtual guitar instruments, and, of course, making things more intuitive for the user (who may or may not have in-depth knowledge of guitar techniques – though I’d think someone who has that in-depth knowledge would probably just play a guitar in the first place).

@Christian_Schormann

You do point out one thing that my prototype does not do yet (thanks!!!), but which should actually be fun to experiment with: Taking a recording (just pure midi without any helpful hints) and finding a default fingering. I just recently found some algorithms that might be helpful in doing this automatically (e.g. GitHub - natecdr/tuttut: Converts a midi file to guitar tabs (or any fretted string instrument) · GitHub ), but I haven’t experimented with that yet. I may try that if I find the time.

My very quick skim through this suggests it is just oriented toward making tablature charts, but maybe I’m missing something. or maybe the making of tablature charts provides an intermediate step that the software can then translate further.

@Christian_Schormann

For chord patterns: I agree that with libraries, the hunting problem is always pretty daunting. I don’t have a good solution for that, but I do think that searching based on a rhythm you play should be quite feasible, which might make a library more useful.

This reminds me of MusicLab’s past support for using some video game (Guitar Hero?) controller that could essentially help convert “air guitar” movements into something their Real series virtual guitars could use for strumming patterns. Not being a gamer (at all), I never had the hardware but always thought it was a cool idea, and maybe I could be a good enough air guitar player that, if I’d had the hardware (I doubt they support it these days, and I’m not sure the hardware is still in production in any case), and it worked well to translate the moves to RealGuitar or RealStrat or whatever, it might be helpful.

But a similar thing could potentially be useful for simply translating chords and rhythms played on a keyboard to a more guitaristic result. The problem with hunting for patterns is that there mostly isn’t a good model. Session Guitarist does at least provide some ability to specify rhythms in its search, though the results seem to be highly mixed. And the other thing with Session Guitarist is you only get 8 (IIRC) patterns in a patch, and the reality is typically needing a lot more than that for a whole song. UJAM has more between their common patterns (something like 16) and the style-specific patterns (8, I think), but there is no rhythmic search or graphic display or …

Toontracks’ Tap2Find is a reasonable model for what could be useful, though, even better would be analyzing a recorded MIDI part and mapping that to strumming patterns over the course of a song – of course, giving the ability to modify it later to finesse it to get closer to a polished result, which is one aspect I particularly like about what I’m reading into your model.

@Christian_Schormann

To your point regarding an instrument: I did not mean to suggest that Steinberg should have an exclusive instrument, but I do think it would be good to have a default instrument to go with the editor (which might introduce a default, “standard” mapping).

Yes, I have no objection to this, and having an instrument at least serves as a teaser to get people interested in the editor facility. It’s just very important that the facility work with mainstream instruments that people will already have, and right out of the shoot.

I don’t know if you’ve kept up on the (many) AI-oriented discussions that have gone on in these forums, but, in the context of this discussion, I would offer a response I made in one of those that gives something of an overview of how I envision AI might be able to help in my personal applications for trying to get more realistic guitar parts from virtual instruments:

AI generated tracks ?!

While I don’t see this as a feature of Cubase itself – it seems to me it would more likely come in plugin form – this is pretty much exactly where I stand on the potential use of AI in this context. For perspective, I am also a keyboard player (and singer and songwriter and …) with a project studio. I am decidedly not a guitar player, strings player, brass player, etc. Over the decades, though, I’ve played and/or programmed all the instrumental parts on all except an extremely small number of …

Of course, it doesn’t necessarily have to be AI to do this – my comment at the link was made in the context of an AI discussion, but I’d be open to any sort of solution that helped on this front, and it might be that some of your ideas here could do that to a degree. It might also be that this could be an area where a combination of AI-based advances and something like your guitar-centric/stringed instrument-centric MIDI editor could make a real difference.

[Christian_Schormann]

@rickpaul

though I’d think someone who has that in-depth knowledge would probably just play a guitar in the first place

Guitar is my primary instrument, but I still often prefer to program guitar parts instead of playing them. Why? A few reasons are:

• In many cases, it would take me longer to play a part with the desired precision and quality than programming it.
• I have better control over the expressivity and details when I program compared to when I play
• I have better control over the sound of the guitar track when I program
• I am on the road, working in a cafe/airplane/hotel room, and I don’t have the equipment to record a life guitar with me

I know some people will find these reasons unqualified, but I also know I am not alone in this. In the end, the reasons to use a synthesized guitar are the same as for any sampled or modeled real instrument: Convenience, availability, flexibility, control, skills (or lack of them)

@rickpaul

My very quick skim through this suggests it is just oriented toward making tablature charts, but maybe I’m missing something. or maybe the making of tablature charts provides an intermediate step that the software can then translate further

Generating tablature from Midi (or MuscXML) note information requires guessing two bits of information: On which string you want to play a note, and on which fret? In many cases, there are multiple options how you could do it. The goal of the cited algorithm is to find a fingering option that is likely to be easily playable, which often is a choice a guitar player would make. So, if you record something on a keyboard, and you have a good algorithm like this to convert your recording into likely good fingerings, you will have much less editing to do to make it sound good.

@rickpaul

I don’t know if you’ve kept up on the (many) AI-oriented discussions that have gone on in these forums, but, in the context of this discussion, I would offer a response I made in one of those that gives something of an overview of how I envision AI might be able to help in my personal applications for trying to get more realistic guitar parts from virtual instruments

AI of course is a bottomless barrel type of topic. I have no doubt that AI will be able, with appropriate training, to generate all sorts of musical tracks. There are already several plugins and services out there that generate midi information with AI, instead of ready-mixed audio (like suno et al).

I am personally very wary of this. At what point do I stop being the composer of a track if I use AI to generate the tracks? When I just do one track? half of them, 3/4 of them? But perhaps more importantly, I think we need to be a lot more specific by what we mean when we say “an AI generated track”. If we mean “I type in some text prompt and an AI generates a track from that”, I find that a lot more problematic that a case where I write or record a midi track, and I then ask AI to modify its fingering or strumming to make it more like what a guitar player might do. The difference, in my perspective, is in the degree of creative control that I retain as a composer. But of course, this is a question that every artist needs to answer for themselves: At what point, when I use AI, do I just commission art to be created by AI, as opposed to creating art myself? I barely have an answer for this for myself, and even that is changing as I experiment, learn and play… So definitively not judging.

An area for AI use in Cubase that I personally would be much more interested in is in mixing. Traditionally, we are looking at mixing by processing tracks individually (by placing EQs, FX etc), but several new plugins show the benefits of analyzing sets of tracks and applying processing in the context of all other tracks. Unfortunately, these plugins are painful to use because you have to insert feeder plugins on all tracks or set up side chains manually, which is just no fun at all.

I would love for Cubase to have a mixing architecture that can do that sort of stuff automatically - analyze multiple tracks and apply processing to automatically carve out frequencies and avoid resonances and level and so on. Obviously, this would provide an editable base for you to make your own decisions, rather than a fully automated mechanism. This of course could also be open for plugins - imagine a VST extension that would allow for multi-track analyzers and multi-track processors that all could work in concert…

[raino]

As an emerging ex-guitarist (thanks father time) I’ve found that AAS Strum has developed an interface that is pretty intuitive to guitarists. Their MIDI keyboard is divided into 3 sections. In one you play the chord you want and the voicing doesn’t matter. Then in another section you pick where on the fretboard to play the chord, which determines the voicing just like on a real guitar. Finally the last section is where you trigger when to pluck each ‘string’. So the first 2 sections manage the fretboard and the 3rd the picking.

[HPDoubleyou]

In general, I see two possible approaches:

1. A dedicated Midi guitar parts editor., as a part of Cubase. The challenge here is covering the combinations of chord based strums / arpeggios plus solo melodies with individual strings, plus the different ways of mapping this with the various guitar plugins on the market. But it can be done.

2. My personal hope is that guitar plugin makers will use the ARA functionality used by Celemony Tonalic (I am not talking of Melodyne here!). Thus, the guitar plugin maker can use their individual editor, tailored already to their plugin functionality plus offer full integration into the Daw for placing events on a visible track in the Daw arrangement view., follow the Chord track and time sig changes of the Daw.

Just my two cents. Cheers!

[Christian_Schormann]

@HPDoubleyou

2. My personal hope is that guitar plugin makers will use the ARA functionality used by Celemony Tonalic (I am not talking of Melodyne here!). Thus, the guitar plugin maker can use their individual editor, tailored already to their plugin functionality plus offer full integration into the Daw for placing events on a visible track in the Daw arrangement view., follow the Chord track and time sig changes of the Daw.

The problem with the ARA functionality is that it is not really intended for MIDI plugins, but for audio tracks. The idea of ARA is to enable an audio processing plugin to handle events on an audio track. As a result, in my opinion, it is not really user friendly because it requires audio events to work, which cannot be created as easily in Cubase as Midi events. I have plugins that try to use ARA in this manner, and I find it quite unusable. Perhaps if Cubase would introduce a way to create audio events that can be consumed by ARA plugins as easily as Midi events on Midi tracks, this might become an option, though.

Two more concerns with ARA: The plugin interface is complex, I think, way more complex than necessary for a midi editor plugin, which makes it less likely to be adopted for this purpose.
Also, I find that the ARA functionality is by far the most crash-prone part of Cubase, perhaps due to the complexity of the functionality.

I completely agree that it would be the best solution if Cubase had a plugin interface that would allow for the creation of custom editors for Midi events on Midi tracks, or something equivalent that would make the user experience cleaner than it is today with ARA plugins for non audio applications.

[Christian_Schormann]

I have used AAS Strum a while ago. It has a decent interface, but as far as I remember it doesn’t address the problem of providing control over where on the fretboard things are played. Because of its modeling approach this may also matter less for the sound in this particular plugin. But I am not sure, I have not used it for too long.

[Christian_Schormann]

I want to re-emphasize the last point from my original post by shamelessly quoting myself:

@Christian_Schormann

I’d also be super-happy with a plugin interface that would allow me to create an editor like this myself and run it inside of Cubase…

It is of course in the nature of specialized editors to only be of interest to a limited audience. That makes them less interesting for a generalist vendor like Steinberg to create. But specialized tools, sometimes custom-made for a single use, also usually provide a huge productivity boost for the artist using them. In other tools in the graphics space (e.g. Blender, Fusion 360), I am creating scripts sometimes just for one task, and it can make a difference between a project being impractical and being possible. Even some plugins now support scripting, for example UVI Flacon or Synthesizer V.

Steinberg knew this truth once - they realized that they could never create all the audio plugins and instruments people would want themselves, and thus created the VST standard, and the ASIO standard as well.
I wish Steinberg would reclaim this mantle of openness and extensibility that drove it in earlier days and open up Cubase for scripting and plugins beyond the areas that are accessible today. Other DAW vendors are doing it - for example Ableton and Reaper.

I don’t want to have to rely on Steinberg for features that others may perceive as esoteric, I want to be able to do them myself. And in the era of AI-supported coding, writing plugins or scripts is far more accessible than it ever was.

[HPDoubleyou]

I do agree with you 100%. Maybe it’s time for VST 4.

[rickpaul]

@Christian_Schormann

In the end, the reasons to use a synthesized guitar are the same as for any sampled or modeled real instrument: Convenience, availability, flexibility, control, skills (or lack of them)

Of course, it is anyone’s call how they achieve their recorded results on this, so I’d never think to argue with you on this. For the recorded quality side of things, I also use virtual pianos instead of my acoustic piano (which is of a sort that no one would actually want to put in a recording, even if I had a good space to record it in), and I can, of course, edit out mistakes and such. However, I can’t imagine ever programming a piano (or any keyboard) part, because I can get in one (or a few takes) a feel that would take me forever to try and approximate via programming. Even when I’m doing wind parts, I use a wind controller for tracking the parts for a similar reason. (I’m not primarily a wind instrument player, though I did play clarinet and sax long ago when I was in school. But the ability to get the expressiveness quickly by actually playing with a wind controller works far better for me than trying to simulate that with any sort of keyboard-based CC, no less drawing curves.) But perhaps guitar-oriented MIDI controllers aren’t as capable as keyboard- and wind-oriented controllers in terms of translating the physical expression to MIDI expression that is usable with virtual instruments?

@Christian_Schormann

There are already several plugins and services out there that generate midi information with AI, instead of ready-mixed audio (like suno et al).

Just to be clear, I’m not suggesting “generating MIDI information with AI”. That I am interested in is more like “transforming MIDI information with AI”. That is, I play something on a keyboard that is intended for use with a virtual guitar. What I’m wanting is for the AI to add the sort of expression (keyswitches, CCs, etc.) to make it feel more like a real guitar player. Of course, I could disagree with some of its choices and edit the results further, but having to add all (or most of) that from scratch is the thing that both takes forever and requires sufficient ability to “think like a guitarist” (which I can do to a degree, but certainly nowhere near to the level that I see MusicLab’s videos doing where they recreate iconic guitar parts with their instruments).

@Christian_Schormann

I am personally very wary of this. At what point do I stop being the composer of a track if I use AI to generate the tracks?

Again, this is not what I’m asking for. I want to play the parts. I don’t want to program them (which would be more of a parallel to asking AI to generate them). The better parallel might be if I’m an “okay” guitarist, so make a demo of my song, then get an excellent guitarist to actually record the part for the final recording, but asking him to play what I play, just making it feel better. I’m still the composer/arranger, just as I would be if notating parts for an orchestra but having real orchestra musicians play them. In this case, though, I also want to play the parts, just get more realistic expressiveness out of them than can be delivered by playing a keyboard controller and without extremely tedious programming.

@Christian_Schormann

I think we need to be a lot more specific by what we mean when we say “an AI generated track”. If we mean “I type in some text prompt and an AI generates a track from that”, I find that a lot more problematic that a case where I write or record a midi track, and I then ask AI to modify its fingering or strumming to make it more like what a guitar player might do.

Agreed, and the latter is exactly what I’m wanting.

@Christian_Schormann

The difference, in my perspective, is in the degree of creative control that I retain as a composer.

Right, and, just to add to that, I also want to be able to modify the results further to suit my needs in areas where what the facility used to make my MIDI performance more realistic veers from what I was hoping for (or where I just want to experiment more to see if I can better approximate what I have in mind). It’s one thing to do all the programming manually, and quite another thing to have the programming tweaks added but just have a few areas where I want even more control.

@Christian_Schormann

At what point, when I use AI, do I just commission art to be created by AI, as opposed to creating art myself?

That’s a totally different discussion in my mind. But it’s not unlike the decision someone makes when doing their own recording but hiring other musicians to play the parts for instruments they don’t play or don’t play well enough for the results they want to achieve. If I could afford to, I’d certainly be inclined to hire real guitar players to play the guitar parts on my recordings. Ditto for most of the other instrumental parts – “my thing” is mainly writing the songs, singing, playing piano (and other keyboards), and coming up with the basic arrangements. I’m getting better at the engineering and mixing sides of things, but, even there, if I had sufficient budget, I might be inclined to hire pros. It tends to take me on the order of a month to make each of my recordings. At that rate, a 12-song album usually takes me on the order of a year. If I could just afford to go to a pro studio with professional engineers and session musicians, I could probably do an album in less than a month.

@Christian_Schormann

An area for AI use in Cubase that I personally would be much more interested in is in mixing. Traditionally, we are looking at mixing by processing tracks individually (by placing EQs, FX etc), but several new plugins show the benefits of analyzing sets of tracks and applying processing in the context of all other tracks. Unfortunately, these plugins are painful to use because you have to insert feeder plugins on all tracks or set up side chains manually, which is just no fun at all.

This is also another topic from the virtual guitars thing, but, FWIW, I do use some of those plugins (e.g. the Waves Curves series), at least to a degree. For example, I use Curves Resolve to help let the vocals cut through the instrumental mix and to help reduce conflicts between the kick and bass. I’ve also tried Curves AQ in a channel strip-like context, and sometimes I’ve even ended up using it in my mixes. I do find they can be helpful at times, and I don’t much mind needing to set up the sidechains where needed (Resolve always needs one or more, while AQ and Equator, which is for reducing resonances, don’t necessarily). But I’d certainly be happy for any mixing assistance some future version of Cubase might provide if it helps make my mixes better and/or quicker to achieve, while still providing a sufficient degree of creative control.

But, based on my uses to date of Curves AQ, which is one of those plugins that seems to get promoted as magic mixing (i.e. play your signal through it, and it gives you a choice of 5 alternative starting points, which you can then sculpt further, at least to a degree), I think I probably need a lot more creative control, or at least more typically get better results from doing my own thing. (I’ve never attempted using that plugin for every track in a mix to date, and, even when I try it, I probably don’t end up using it more frequently than I do.)

[Michael_Ill]

I would love to see more guitar features in Cubase, but unfortunately guitar players who prefer to edit notes with MIDI are fairly few. I am one of them and also an avid user of MusicLab products, and I also prefer to do programming in the host, not the plugin.

A chord librarian might be something to combine with the Chord Track/Chord Pads at some point if they ever decide to revamp that section. With so many different voicing of guitar instruments it’s easy to forget the exact fingering used. - And in my view Chord’s section really needs a major overhaul.

TBH I don’t even know of a stand-alone guitar chord librarian that does what I want. I have written them all down on paper and I would love to transfer them to some kind of app at some point. If I was to lose them, it would be a small tragedy.

As far as a tab editor to edit notes, I am happy enough with the Key Editor.

[rickpaul]

@Michael_Ill

As far as a tab editor to edit notes, I am happy enough with the Key Editor.

I will say, that I think the tab editor idea could be helpful to instantly see which strings are being played in any given case. As a non-guitar player, who knows enough about guitar technique (even if not being able to actually play) to consider things like strumming only some of the strings, I have to think a lot about what notes would be on which strings when just doing this in the key editor. Whereas being able to see the notes directly on the strings would make it obvious.

[MrSoundman]

It has been mentioned already, but for an example of what a really good guitar MIDI editor might look like, Ample Guitar have recently updated their free “Lite” versions “Ample Guitar M Lite 4” and “Ample Guitar P Lite 4” (a Martin and a Precision respectively). They can be downloaded and installed without requiring sign-up or activation (link here).

If you just want to have a look at some screenshots of the “Riffer” editors built into the plug-in, click here and select the Gallery tab.

[Christian_Schormann]

I have several of the Ample products - these are good products. The core problem with their editors, in my opinion, is that they are not well integrated into the DAW (and that is not their fault - Cubase does not allow that, simply). The workflow, independent of the editor features themselves, is not functional for me.

I think it is interesting that so many VST instrument plugins feel the need to implement custom editors - it shows that there are needs that are not met by what is available inside the DAW. Some just let you work on a pattern which makes it very hard to work and iterate on an arrangement level. Some, like ToonTrack or Synthesizer V, provide their own arrangement timeline where patterns can be arranged, but that now leads to a situation where you may have juggle multiple timelines with they own set of clips, so that any global edit to the form of a song has to happen in multiple interfaces instead of one place.

I think this is quickly becoming a mess. I would very much prefer a situation where the synthesis of sound and the editing of the input for said sound remain separated, so that the DAW timeline is the place where all arrangement happens, but where custom editor are available for content editing tasks for the clips arranged on this timeline.

Cubase does of course understand it - that is the reason why we have a key editor, a score editor, a drum editor and a list editor. I just want more such editors, or even better, in an ideal world, the ability to write my own.

For audio tracks, Cubase already gives us this capability via ARA, but this is, in my opinion, not the right workflow for midi-based tracks and instruments. I’ve don’t even think it is realistic for Cubase to answer all sorts of specialized needs for editors - in my mind, Cubase is a platform that should focus on the mainstream use cases but provide strong extensibility for special use cases. This extensibility is missing today.

[Christian_Schormann]

@Michael_Ill

As far as a tab editor to edit notes, I am happy enough with the Key Editor.

The problem I have with the key editor is not necessarily just entering the notes, but:

• Entering the auxiliary information that causes a guitar VST instrument to reproduce the notes on the desired string/fret combination. Depending on the instrument that requires either extra keyswitch events with precise timing requirements, and/or a distribution of events across multiple channels, all of which are slow and painful to enter and hard to maintain as you edit and iterate
• Entering even more auxiliary information to get the right guitar-specific expressions
• Handling chords and strumming efficiently
• Visualizing and understanding all of this, when you look at something you did two month ago and are trying to edit it then

Don’t get me wrong, all of this can of course be done in the key editor - I do it all the time, but it is just so slow and painful. I’d like a better solution for it.

And again, I completely understand that this may be specific to my needs, or the needs of a small group of people. That is why I said from the beginning that I don’t expect Steinberg to build something like this, I would just like them to give all of us the extensibility interfaces that make it possible for us to build and integrate our own specialized tools.

[emotive]

I use Ample Guitars, and for what I do, they are the best for guitar programming in the Key Editor, since they let you know when specific note combinations are actually able to be played by a guitarist, otherwise notes that are impossible to be played, are muted.

I avoid custom plug-in editors wherever possible. For example, I use EmvoiceOne, for vocals and it is brilliant at what it does, but notation is done inside the plug-in, and while it works with variable tempo, you need to feed it new tempo information prior to modifying existing notes, when opening a project so most of the time the plug-in is disabled since it uses the network, and everyone knows what that can do to the stability of a DAW.

Having editors available that are specific to instrument types, such as stringed instruments, starting with a standard guitar, in Steinberg products, would be a differentiating feature so I can’t imagine that Yamaha and Steinberg would not be interested since they are the only developers with a Drum Editor but would a guitar editor need to be based on .xml files?

[raino]

A bit off topic, but I’d like to see a MIDI Controller structured like an autoharp. One hand would generate strumming patterns while the other selects chords.