Hey. I’m still searching for that magical electric lead guitar virtual instrument that I can use to create truly nuanced (and REAL sounding) lead guitar lines.
As a reference point, I use the SWAM saxophone modeled virtual instrument, and it lets me get every aspect of a sax solo down JUST the way I want it (breathiness, trills, vibrato, attack, growl, “expression,” you-name-it). I’m mystified as to why there don’t seem to be any similarly sophisticated lead guitar virtual instruments that offer the same kind of nuanced realism (and/or emulation of specific artists and/or styles of playing).
Even some of the poughkeepsie AI generated guitar solos I hear from platforms like Suno sound MUCH more real and textured (and even vibrant) than what I seem to be able to get out of the virtual guitar instruments I’m familiar with.
So my question is: what are your go-to lead guitar virtual instruments when you want to cut a super-realistic sounding lead electric guitar line? [For instance, I’d like to be able to cut “Jeff Beck level” nuance and range (and power) into some guitar solos I have in mind – does anybody know of an electric guitar VST that can do this, or have any tips on how to get this kind of realism from ANY virtual electric guitar instrument?]
As weird as that may sound, none of the instruments SWAM does are even close to the level of variants or complexity a guitar has. Going for your Jeff Beck example, the way he plays is already outside of what pretty much every single guitar VSTi in the market does. You can’t sample that kind of playing and synthesizing it represents about as much challenge as trying to sample it. As an example, on a single phrase, you can do different articulations on each note. Maybe the first is picked, you hammer the second, the third is pulled off and the last is pre-bent. A MPE-capable controller in the hands of someone who knows what they’re doing might get close, but not quite, as shown here:
The point here is you’ll have to reach some level of compromise with whatever plugin or combination of plugin you decide to use, as you’re not gonna get them to sound exactly like a real guitar. You might get close enough to the point most people won’t be able to pick it up at first.
In my opinion, Ample Sound Guitars are the most realistic VSTs if they’re well programmed. A guitarist’s help can be useful for programming, as it needs to take into account a guitarist’s natural playing style.
@BrunoDSL’ s notes above are definitely on the money about the complexity versus wind instruments. (I’m still using the SampleModeling saxes and trumpet and trombone with a wind controller for those.) I have to say the Roli versus guitar video was impressive, though I have to wonder what specific virtual guitar he was using and what was involved in mapping it to the Roli keyboards controls.
As for the original question, I’m not sure how realistic I get in terms of my results (better than some, not as good as others), but I’ve used a number of different virtual guitars for leads, including all the electrics from MusicLab, Acoustic Samples’ Telematic V3 and Strategy, the melody versions of NI Session Guitarist electrics, and Chris Hein Guitars (though I haven’t used that all that much recently). Which one I use depends on the musical style. For example, if it’s along country lines, odds are it will be one of the Tele options (Telematic or Electric Vintage). I tend to find the Acoustic Samples instruments map to just my playing (on a keyboard controller) most intuitively. I’ll typically do multiple passes, with the first being playing the notes, then overdubbing mod wheel and pitchbend. Those don’t provide a lot of custom CC enhancements, so they’re probably not going to get the really nuanced stuff.
If a lot more nuance is needed, the MusicLab guitars – I probably use RealLPC the most – have a lot of programming capabilities with keyswitches and CCs (and they give flexibility in assignment). But, aside from setting up some of your most needed (and playable) bits, which may provide at least some enhanced realism for the initial pass, we’re likely talking a lot of programming of CCs. For example, here’s a tutorial on using RealLPC to create the guitar solo in Bon Jovi’s “Bed of Roses”:
That is WAY more editing that I’ve ever wanted to do!!! My main thing is typically play a few rounds of takes, comp, overdub playable CCs (if I’m just doing single note leads, I may have played the pitch/mod lever when I was tracking the notes), then, only if absolutely needed to get the results I’m going for add in additional CCs and/or keyswitches.
The other key part is, of course, the processing on the electric guitars – i.e. pedalboard, amp and speaker simulators, etc. While the MusicLab stuff now has fair capabilities built in on that front, I really haven’t used those much, unless one of their presets (e.g. from the Top Guitarist series) just happens to work. And most other virtual guitars I’ve used haven’t had the flexibility I want on that front. My new favorite on this front is the UAD Paradise Guitar Studio, which just came out early this month. Before that came on the scene, my most frequent plugin for that was NI Guitar Rig, but I’ve also used some of the Waves stuff (e.g. the PRS series and Voltage Amps, as well as their pedalboards), Overloud TH3, and IK AmpliTube. I’m usually auditioning presets before I even track the lead guitar parts because the sounds influence the way I play when tracking, and that can ultimately make a big difference in expressiveness.
That’s probably Roli’s Equator synth. Back when the seaboard came into the market, MPE was still a new thing with the closest thing being the Haaken Continuum, which was not polyphonic. Nowadays, you have free MPE capable synths like SurgeXT.
In looking at Equator just now, I wasn’t really seeing anything suggesting it did “real” instruments (other than drums) – it seemed like it was more for programming creative combinations of sounds – but you may be right.
That led me, though, to searching on the topic of virtual guitars and Roli, and I came across a few threads here and there that mentioned a guy named Marco Parisi, specifically mentioning his performance of “Purple Rain” (and also “Little Wing”) that way. Here’s a link to “Purple Rain” – skip to around 7:14 for the lead guitar solo:
People in those threads were asking what virtual guitar he was using, and some said he kept it highly secretive. However, one of the threads also had a link to a short video (I must have already closed the window where I could have gotten the link), where he claimed to have used a Jazz Guitar patch from Kontakt’s Factory Library through Guitar Rig Pro. My curiosity on that also led to some videos where people were configuring some other Kontakt-based guitar libraries for Roli keyboard use, specifically changing pitch bend ranges to an octave and sometimes using six instances of the same library on different MIDI channels to be able to do per-string expression.
Part of my curiosity here is that the Roli performance you linked felt like, in the right hands (i.e. after a learning curve), using one of those keyboards might really have the potential to raise the level of virtual guitar playing realism, and I noticed there is a Seaboard M currently under $300 (and usually only about $50 more expensive). Having just received a little unexpected Christmas money, I’m at least toying with the idea of whether it might be give me more mileage than some of the other products I’ve been considering. (Beyond just covering a 2-octave range, I’m also concerned about the miniature keys given I’m mainly a piano player who’s played full-size keys for all my life.)
Well, for the same price of the Seaboard 2, you can get a Linnstrument. It’s not piano shaped, but it’s open source hardware, class compliant and all the configurations of it are done on the device itself, so no proprietary software or drivers are necessary. Roger Linn was also nice enough to make some MPE capable presets for SurgeXT and those come with the plugin for free:
Thank you for the video featuring Marco Parisi - it’s impressive what he does. However… it’s still pretty far away from a real guitar solo - there are just to many tiny things that change with every single note and these make all the difference.
It’s different and good in its own way. But it’s surely not the real thing.
Somehow I have this mental image of a spray-painted Thanksgiving tofu turkey in my head…
LOL. Well, yes, of course, it’s not the real thing. Nor are the orchestras done from virtual instrument libraries, however convincing the person using them may make them sound.
For me personally, not being a guitarist, I’ve used many different products to simulate guitars over the decades, starting with MIDI module patches back in the 80s. Over the years, as things made things better and better for purposes of making things more realistic, I’ve added or switched tools as made sense to try and raise my game. For picked acoustic guitar parts, the current state of the art (for me, especially, the Acoustic Samples GD-6 and Sunbird), I’ve gotten to the point where I’ve gotten compliments on the guitars from some trained reviewers (e.g. screening for music libraries), just playing, with no programming.
Strummed rhythm guitars had always been more challenging, but, starting with the original Steinberg Virtual Guitarist, which combined sampled real guitar parts with chord detection, those started getting easier IF the patterns I needed fit what the virtual instruments provided. The modern descendants of that product (the UJAM Virtual Guitarist series) provide more flexibility on that front, though there’s still a good deal of programming (especially on the key switching front) involved. But those parts also tend to be more background in the context of a song. (The MusicLab and Acoustic Samples products also provide strumming options, as well as some pattern play options, but I find those less friendly for strumming, even though I’ve used them for that here and there.)
With the Marco Parisi thing, though, as well as the other Roli video posted in this thread, the key thing that struck me was that they were playing this in real time, not going back and editing what they played in real time later to add nuances. I assume that, in the context of a recording, that could also be done IF what you played in real time didn’t measure up to what you want for a specific context.
Of course, one other thing that comes into play is being able to think like a guitarist to have a sense of what you want, and know what to program, if you’re going in that direction. But that’s no different from virtual drumming or virtual wind or string instrument playing, even if it might be more complex. (I still have not managed to get fiddle parts to a point of believability that satisfies me, and I think pedal steel may actually be the world’s toughest instrument to handle in that way, even more so than electric guitars.)
I agree - it’s all a matter of context. What I was trying to say: sometimes it’s better to embrace the imperfections and focus on the things a virtual instrument is really good at. It’s different - that doesn’t necessarily mean it’s bad. And some strummed guitars sound pretty decent if put in the right context!
@Googly_Smythe If we start to talk about amps this thread is likely to become one of the longest this forum has ever seen. Do we want that…? Probably maybe…
For some reason I only saw this days later. It was a long, slow-paced video. While the hardware decidedly wouldn’t work for me (I am decidedly a keyboard player, and not even with things like the left-hand part of an accordion – beyond that, I can handle percussion and wind controller with clarinet/sax-type fingering), as I was skipping through the video, just checking out the various sounds he was featuring, I will at least say his “okay pedal steel” example was actually pretty impressive on the believability with the combination of the sound itself and his manipulation of it with the controller.
I often recommend the Linnstrument because of the things I mentioned and the fact it’s not designed to be disposable like the ROLI Seaboard. You’re not gonna see many ROLI units for sale second hand because the first thing to go is the rubber surface and ROLI doesn’t replace those. If you want to try that repair yourself, you’ll be “happy” to discover ROLI uses Apple’s pentalobe screws in their stuff and a lot of the parts are glued.
Understood, but, for me, that’s equivalent to recommending I buy a guitar because it’ll last longer than an electronic keyboard. Probably true (you don’t see many 1950- or 1960-something electronic keyboards having meaningful value), but, not being able to play guitar worth beans, having one wouldn’t do me any good. (The instrument I play most often, albeit not on recordings, is the 1940-something Story & Clark spinet that was originally my mother’s when she was growing up and was the piano I learned to play on. She got another piano when I was in high school, and the old one went in our basement practice room, but she gave it to me once I moved out permanently.)
That could well be a concern if I decide to go for one, but a least the Seaboard M (only version I could consider at present for financial reasons) is under $300. Also, from what I read on the various discussion forums when I was looking into this, there was a fair mention of Roli having upgraded their quality on this front in the more recent models. Nevertheless, it’s easy to see how the flexible rubber surfaces could have wear and tear issues, especially if used a lot. (Hard to say how much I’d use it if I go that way. I think it would be more than my wind controller since I have guitars in the vast majority of my recordings, but winds only occasionally. But, for desk space considerations, as well as control surface considerations, I probably would only hook it up when I need it, as opposed to making replacing my Roland A-800Pro as the main controller on my desk – basically for everything except actual piano parts and winds.)
Yeah, that part “describes” me similarly to how being in the market for a guitar controller would. (I’m definitely NOT handy on any sort of repairs.)
My go-to lead guitar is me. On the one hand, I appreciate that of all instruments, lead guitar is the hardest to sample convincingly. Guitar lead lines are nuanced because real people make the nuances. Compared to other instruments, we can bend strings, vary our vibrato, hammer on, pull off, and with all those techniques, we can make a lead line sing. And these techniques are exploited in every solo. I play in two different bands and solo extensively.
On the other hand, why is everybody in such a hurry to replace lead guitar lines with a virtual instrument in their music? I think if you don’t play guitar, or you’re not willing to hire a guitarist, even one who would probably play for free, then why not just use a different instrument? Why are you trying to get a sampler to generate nuance? Just generate your entire song with AI.
Because they don’t want to. They want a guitar. Just like if someone wanted to write and perform an entire orchestral score when they can’t hire a full orchestra.
Because some are quite good at it? Parisi’s performance of Purple Rain on the Seaboard is a great example, but maybe you didn’t read through the thread?
I think you misunderstood something, or you’re just being flippant on purpose. There’s absolutely nothing wrong with a keyboard player wanting to use sample-based guitar instrumentation which can provide nuance and humanized styling if that’s what they want to do. It’s great you play guitar! . Some people don’t, yet still choose to write and compose their own original songs. There’s an entire universe which exists between “finding someone to play guitar for free” and “writing your entire song with AI.”
Well, I don’t play guitar worth beans (and don’t even have a guitar around), and I can’t afford to hire a guitarist – I wouldn’t be able to put out anywhere near as much music as I make (and I’m not super-prolific on that front). The caliber of guitarists I’d need if getting someone to do what I want aren’t likely to play for free, nor would I want to impose on them for that. And there would still be the need of having legal agreements in place for commercial use of the results.
In general, though, I also know what I want, and I play it. Over time, the tools (and my skills from learning to apply them) have allowed me to get closer and closer to the sound I hear in my head. While I do use sounds other than guitars, a lot of my music, just stylistically/genre-wise, is guitar-based – especially pop rock, soft rock, and country – and, when I want a guitar lead, I want a guitar sound, with at least a level of nuances that allow “suspension of disbelief”. Maybe a trained guitarist might pick up that it’s not a guitar, but the average layperson would hear a guitar. (Sometimes even trained evaluators do as I’ve gotten compliments on my “guitar work”, especially in picked acoustic guitar contexts, on some of my recordings.)
As for AI, while I’d never use it to “generate (an) entire song” – I am a songwriter, musician, singer, arranger, etc., and why would I want to give up the activity I enjoy just to get some potentially more believable result? – I have had a little recent exposure to AI through one of my songwriting collaborators’ efforts. He’s been looking at some of his old songs, including ones we wrote back in the late 90s, and is considering the idea of using Suno recordings to put them out. (He’s already got one song out there, not one of our collaborative ones, and I have to say, it was an excellent performance of a solid song.)
In experimenting with one of our songs, he initially converted an old RealAudio piano/vocal demo I’d done to some form Suno would accept (before I sent him a WAV file and reminded him he actually had it on CD) and had Suno make a female country recording from that. It was pretty good (and surprisingly emotional) but made some key errors. So I worked with him a bit in terms of feedback, similar to how a producer might work with an artist and studio musicians.
The thing is, though, at least with the Suno Pro plan he has (not sure if their high end plan that includes Suno Studio would have helped on this front), each attempt at making a change is a “do over” (i.e. from scratch), as opposed to the sort of incremental change that would be made when working with real musicians or just doing the recordings myself and going from initial arrangement through work mixes, rough mixes, and final mixes in refining my recording. It was pretty frustrating because there were very good things in each version, but there were also very “wrong” things (e.g. messing up the vocal melody, chord changes that jarred, etc.). Some versions were better than others, but none were acceptable as is.
Ultimately, I suggested he send me stems for the various versions, and I reviewed all those. They were all at the same tempo, but some had different structures in terms of things like instrumental breaks. I ended up selecting the version that had the best instrumental arrangement, then effectively comping the lead from parts of all the other versions (thankfully, the lead vocal sound was pretty consistent between versions) to be able to pick and choose phrases (and sometimes parts of phrases). There was one part where I ended up having to use Melodyne to change notes because none of the takes had that part right on both the phrasing and melody notes front. There was also a fair bit of fiddling to make the parts from different versions hold together, then to make the composite vocal fit back in the instrumental mix, but it ultimately turned out quite good.
In most cases, I’d just have done the vocal myself, but this specific song was one where it simply doesn’t work for a male singer (it’s a mother/daughter thing). I can certainly see the value of this recording as an excellent song demo, and the pedal steel (and maybe a bit of fiddle) on it would be something I’d have had a really hard time doing anywhere as well as Suno did. But it also wasn’t anywhere near as creatively fulfilling as arranging it, playing (and sometimes doing a bit of programming to enhance) the virtual instrument parts, etc.
That said, if I could play a virtual guitar part, and get AI to “just make it more realistic”, rather than having to add whatever nuances the specific virtual instrument I’m using provides via overdubbing or editing CCs and key switches, I’d be all for it. But I’m generally not looking for AI to make up its own part.
sorry, that was evil - I didn’t mean it…
. Some people don’t, yet still choose to write and compose their own original songs. There’s an entire universe which exists between “finding someone to play guitar for free” and “writing your entire song with AI.”
But I’m generally not looking for AI to make up its own part.