Next to the guitar itself, few subjects inspire as much passion among guitarists as amps. There’s something about the sound of an overdriven speaker pushing air that you just can’t get any other way.
Or can you?
I used to play only through tube amps, but what I heard in my head also required things that tubes couldn’t do. Although tubes gave a great sound, there was always a certain coloration I couldn’t subtract. I figured I’d go for a neutral amplification system, and add a tube preamp if I wanted that sound . . . right?
Wrong. What makes a guitar amp special is a zillion factors. So, when amp modeling came along that promised to nail these variables, I was intrigued.
There’s no question that for a huge number of pros, modeling does do the job. But are we really there yet? Can you ditch your amp, go with a plug-in, and duplicate the amp experience? We interviewed some of the major players behind amp modeling and plug-ins to share their thoughts — misconceptions, where we need to go next, whether native processing does the job, and more.
GOT LIVE IF YOU WANT IT . . . MAYBE
First up: Can amp plug-ins hack it for live performance? Davide Barbi, head of R&D for IK Multimedia (makers of the AmpliTube series; see Figure 1), says, “Recently we’ve seen a ‘bump’ of guitar players using AmpliTube Live in a laptop, especially for small club gigs. But, with something like AmpliTube 2 that models a huge variety of amps and stomp boxes, it’s reaching the point where it’s not practical to carry around that amount of physical gear. For that level of versatility, economics virtually dictates using plug-ins live. But plug-ins will always remain a very popular studio piece, for reasons like easy re-amping.”
Sascha Kubiak, Head of Product Management at Native Instruments and responsible for Guitar Rig (Figure 2) and Guitar Combos, still sees plug-ins used mainly in the studio. “But increasingly, people are using Guitar Rig on stage — for example, the upcoming live tours of U2 and Megadeth/Fear Factory, where Christian from Fear Factory plans to use it live. We put a lot of emphasis on Guitar Rig’s real-time MIDI control, which we feel is essential for expressive live performance.”
However, Joe Bryan (VP Engineering of Universal Audio, whose Nigel guitar amp in Figure 3 runs on their UAD-1 DSP card) points out some of the limitations. “Suitability for live use depends on the system’s audio through-latency (analog-to-analog signal delay through the digital converters and processing). For some, 5–10 milliseconds isn’t too bad, but getting a DAW to perform well with ultra-low latency remains a challenge. So, a lot of people use amps and DIs for tracking, then reprocess the direct and miked signals later through the simulators for additional control during mixing.
“But for live stage work, nothing beats a real amp. It sounds and looks better, is easier to set up and control, has no added latency, and doesn’t have fiddley wires or connectors to break when you do that stage dive. And, if someone spills a beer on your amp, it just smokes a bit; spill it on your laptop, and you’re playing air guitar.”
Patrick O’Connor, Chief Product Designer for Line 6, definitely sees modeling moving into the live arena, although not necessarily with native plug-ins. “Amp Farm (Figure 4) supports Digidesign’s new Venue live sound console. It’ll be used live with in-ear monitors, in much the same way PODxt has been popular with people using in-ear monitoring live. Todd Rundgren’s guitarist used GuitarPort live for Todd’s last tour; Todd himself used a PODxt. Modeling has the potential to deliver tone-wise for live use, but that tone has to be part of a low-latency, accessible system.”
NATIVE VS. HARDWARE
O’Connor isn’t convinced that native is always the right answer. “Currently available native plug-ins give some great sounds, but consider that many users rely on multiple tracks, plug-ins, and soft synths running simultaneously. Unless they’re savvy about minimizing latency, and sequencer functions like track freeze are implemented well enough not to blow the guitarist’s creative flow, the latency and inconvenience can be a deal-breaker. One reason for the enduring popularity of hardware processors is that they provide processing and low-latency monitoring regardless of the recording applications’s buffer size.”
Bryan agrees that hardware has advantages. “A/D and D/A converters still have millisecond-plus latencies, and small audio buffer sizes require more system resources than larger buffers. It’s not just about plug-ins and tracks; guitarists who like higher sample rates invite bigger buffer sizes. DSP cards can help offload heavily loaded systems, which allows using smaller buffers to reduce through-latency for tracking.”
Not surprisingly, given a company name like “Native Instruments,” Kubiak has other thoughts. “A true component-modeling-based amp emulation like Guitar Rig would not be possible without the computers we have now. Dedicated hardware has no advantages as far as the quality of sound is concerned; the processing power of today’s computer processors is actually superior to typical DSP chips or hardware units. That’s why it’s possible to tweak extremely detailed amp settings in Guitar Rig, like the tube grid bias or the power supply voltage. The emulation algorithms have to be that complex to act with the same dynamic behavior as their real-world counterparts. But really, the most important factor that drives the sound is the quality of the software.”
Barbi agrees. “The current crop of computers are absolutely good enough; you can run several AmpliTubes on even average computers. Besides, the ultimate results depend more on how the product was made, what modeling techniques were used, and whose ears were judging the sounds during development. And a lot of the user experience depends on product concept — it’s important for guitarists to feel like they are in a virtual reality world that gives them what they expect in a simple, easy to use setup.
“However, dedicated hardware is still needed for near zero-latency performance. That’s why AmpliTube 2 will also be TDM — it has less than 1ms delay, which is, for all practical purposes, realtime operation.”
Personally, I used to find playing guitar in real time through plug-ins unbearable. But with my system now hitting latencies of between 3–6ms, that’s definitely more than good enough for a satisfying, “really-close-to-realtime” playing experience.
WHERE WE’VE BEEN, WHERE WE’RE GOING
So what’s the hardest part about modeling? O’Connor’s answer is revealing: “Honestly, the most difficult thing is having the finances, time, commitment, and quality focus that are required to really do it right. It’s taken us eight years to achieve the level of emulation that we offer today, with resources that come only from selling almost $60 million worth of equipment last year. We also had the unique benefit that POD and Amp Farm opened the door to the top guitarists, engineers, and producers. We needed all their advice, as well as the expertise we’ve developed over the years, to achieve what we have with amp emulation.
“But as far as the particular bits of emulation technology — tubes, speakers, etc. — they’re all hard! Our initial years of R&D concentrated on the tubes, then the full amp electronics, with a lesser focus on speaker/mic emulation. Next we focused on deepening the amp emulation detail, and the range of amp tones we could nail. Our most complete efforts in speaker/cab/mic emulation came only with our latest generation products (PODxt, GuitarPort, Vetta II, and Amp Farm 3.0).”
Barbi also stresses the benefits of getting into the game early. “To do it right at every stage is all difficult, because every system element has its own quirks. But, we’ve had many years to perfect our methods and have developed new techniques to best emulate tubes, speakers, cabinets, and mic modeling. Some of this involves analyzing the schematics down to the component level; this detailed analysis occurs before the DSP model study actually starts. And the company consists mostly of musicians who use the products we make, so we have a personal stake in making it right.”
Kubiak notes that the reason why tubes are difficult to emulate isn’t so much the tube itself, but “because the way they distort undergoes subtle changes according to the dynamics of the signal and the load on the amp. You have to treat amp simulations more as a musical instrument than a ‘signal processor’; as convincing emulations come only from meticulous attention to detail.”
Bryan concurs: “Most ‘tube simulators’ just emulate the non-linearity of a single tube, but amps are much more complicated than that. There are several tubes, plus EQ and transformers, each with their own non-linearities, all wrapped with feedback. When this complex circuit is coupled with a speaker cab, it gets even more interesting because the speaker’s dynamic impedance has a huge effect on the amp’s non-linear response. Modeling transformers and how they affect the rest of the circuit is a pretty difficult task.”
THE ACID TEST
Although some think that amp modeling companies put amps in anechoic chambers and crunch numbers to do emulations, that’s not the whole story. Barbi is adamant that “Listening tests with great guitar players is absolutely essential, and can make the difference between a product that just looks good on paper compared to a product that is going to have the character and response that a guitarist expects. This is a very delicate process, and we take it really seriously.
“After the initial instrumental comparisons, we fine tune the models with the help of musicians playing music (not just “testing”), most of the time together with other instruments in real mixes. It’s time consuming, but we think this process is why users tell us that AmpliTube sounds ‘musical.’”
Kubiak has a similar philosophy. “The emulation algorithms are compared to the original hardware in every way, from analytic measurements using test signals to extensive listening sessions with our ‘expert group’ of professional guitarists and producers.” Bryan also sees evaluation as a two-part process: “Analytical tests match the system’s response, but listening tests are always used for final qualification.”
O’Connor adds, “The most important, and difficult, part is finding an absolutely incredible example of the amp we want to model. If you model a mediocre amp, you get a mediocre emulation. For example, the AC 30 we modeled (and now own) was in use weeks before by both Lenny Kravitz and Brian May for their album recordings. Once our model is complete, we A–B it to this “Holy Grail” original, in real world situations with guitarists, until they’re matched. We also probe the original’s circuits electronically and compare them to our model, using audio analysis equipment.”
It’s refreshing that all four manufacturers represented here keep coming back to the sound and feel, because frankly, that’s the only way that plug-ins will ever be able to compete with the amps that guitar players love. But zooming out, one thing is clear: Guitar amp emulations have improved dramatically, as technology has improved and the designers’ ability to bend digits to their will has matured. Even those who aren’t convinced we’ve reached perfection generally recognize that it’s only a matter of time before we come so close any differences will be insignificant. And for many, we’re already there.