Samplemodeling: The Future of Virtual Wind Instruments?

April 12, 2013
share

imgPROS

Amazingly realistic and expressive solo wind instruments. Easier to get realistic performances live than anything else we’ve tried.

CONS

CPU-intensive. Too few presets per instrument. 

Bottom Line

Samplemodeling has created nothing less than a new standard in realism and expressiveness for solo brass and wind instruments. This is the future.

The Trumpet, The Trombone, French Horn & Tuba: $194 each
Ms. Sax S: $155 | The Sax Brothers: $311 | larger bundles available at discount*

samplemodeling.com

*Prices are for direct downloads and based on conversion from Euros at press time.

Sampling has progressed by leaps and bounds since the ’90s, allowing ever more realistic instrument emulations. Most real instruments have been well served by this progress, but a few have been difficult to recreate convincingly via sampling, and attempts at physically modeling them have fallen short of complete realism as well. At the top of that list is the sax family, followed by brass. Sampled sections are doing well, but the lone, exposed sax, trumpet, or trombone still reveals signs of being digital. Formed by renowned developer Peter Siedlaczek and Giorgio Tommasini (a cardiologist by day who has authored a number of papers about improving the realism of samples), Samplemodeling aims to change all that via advanced combining of sampling and modeling. Their product family at the time of this review includes The Trumpet, The Trombone, Mr. Sax A, Mr. Sax T, and Mr. Sax B (alto, tenor, and baritone, bundled as The Sax Brothers), Ms. Sax S (soprano), and French Horn & Tuba.


The Problem

Most sampling engines is that while we can switch between many layers of recorded samples at different velocities, we can’t smoothly crossfade sustained notes in an overlapping fashion without phase artifacts and harmonic incongruities. This isn’t a problem for struck and plucked instruments, but in any wind instrument it means that we can’t fully recreate what happens as you increase and decrease the wind stream of a held note. A lot more changes timbrally than just amplitude, and synth filters can’t realistically mimic these changes. Combine this with the fact that seemingly each note played on a sax sounds slightly different due to breath, embouchure, bite, fingering, and a host of other aspects, and you’re at the heart of why these instruments are so elusive to reproduce.


The Technology

 
 

Fig. 1. Mr. Sax T hosted in Kontakt (Kontakt Player also works.) The C next to a number denotes the MIDI CC controlling the parameter.

Samplemodeling’s Harmonic Alignment technology can crossfade from pp to ff, and from no vibrato to vibrato, with continuity and without phase artifacts. All Samplemodeling instruments start with samples of every chromatic note of an instrument, recorded in an anechoic chamber at every dynamic and articulation. Samples are 24-bit, unlooped, and have a minimum duration of eight to ten seconds. From there, a number of adaptive modeling techniques are applied to capture what the company calls “the performance fingerprints of the real instrument.” These include, but aren’t limited to, convolution, advanced MIDI processing, and “modal resonance” modeling the standing frequencies of an instrument that change as different valves are opened and closed.

For the sax models, a technique the company calls Synchronous Wave Triggering (developed by Stefano Lucato and Emanuele Parravicini) continuously interpolates among different vectors such as time, dynamics, pitch, and formant (vowel-like character of the sound). This allows things like constant-formant pitchbends, mixable subharmonics, and growl and flutter tongue techniques to be performed in real time.


 

Starting To Play

 
 

Fig.2. Controller pages from three different Samplemodeling brass instruments, showing some of the available nuances and articulations.

When you first load a Samplemodeling instrument, it tells you it needs to receive MIDI CC 11 (expression) before the instrument will play properly, even from the onscreen keyboard. So you’ll also want a keyboard controller that has an input for an expression pedal—and that has pitch-bend and modulation controls and can shift octaves (or has at least 61 keys to avoid the need to). I understand the need for expression, as that’s the “wind” beneath the wings of your playing, but I always encountered that speed bump every time I loaded a new instrument or family. It’s easy enough to move your expression pedal, but I’d prefer to have to do it only once per session.

I then played a note on the Mr. Sax T. (tenor sax) instrument and moved the expression pedal. Sold! The way the sound could swell from a barely-played airy tone into a full-bodied bright timbre, then all the way to a “just before overblown” character by only sweeping the pedal was amazing. It took me a little while to get my playing to sound natural, as my first inclination was to accent each note with a pedal move, as with a wah-wah, rather than sweeping the pedal with the arc of a musical phrase or pushing a held note forward and then back. Not to mention adding vibrato to highlight held notes and ends of phrases. But after a short time, I got the hang of it.

Playing The Trumpet was equally inspiring. I appreciated how the vibrato seamlessly morphed from subtle and tasteful vibrato into a “shake” where you hear a distinct alternation between two pitches. The Trombone went from warm and murky to a powerful, “spitty” tone, and Mr. Sax B. (baritone sax) had me moving from Gerry Mulligan jazz lines to Doc Kupka funk licks with only a few parameter changes.

Each Samplemodeling title gave me the same experience. Never before had I tried solo sax and brass sounds that came to life so naturally. I never felt like I was obviously switching a sample by touch, and I never felt like a filter was opening and closing on the same sampled sound. It had the fluid characteristic of playing a physically modeled instrument, but it sounded so right that I knew it had the accurate “snapshot” quality that sampling delivers—only with the snapshots transforming into each other in a way I’ve never experienced. 


 

 
Parameters

When hosted in Kontakt or Kontakt Player, Samplemodeling instruments offer pretty much the same parameters for determining how your playing interacts with the sound, so let’s take a tour through the Mr. Sax T tenor sax (see Figure 1).

In the center are meters showing realtime velocity, expression level, vibrato depth, vibrato rate, and dynamic level. Below these are the settings for the various parameters the engine offers. These include: Master Tuning, Random Expression (a slight varying of the expression value for a more human effect), Controller Mode (keyboard, keyboard with breath controller, and two forms of wind controller), and a Dynamics-to-Pitch parameter, which slightly varies the pitch based on your expression level (harder blowing usually moves the pitch flatter).

The third column of settings contains macros for how the attack is shaped: Enhanced Attack Hard, Smooth Attack Hard, Enhanced Attack Soft, and Smooth Attack Soft. These work by reading the velocity level of your new note and momentarily taking over from the current expression value (i.e., where your pedal or breath controller is at) to produce different response shapes for the newly-played note. The note dynamic starts based on your played velocity and then ramps up or down to the current expression level, producing slight swells, or sforzandos, in a very musical fashion. The parameter below that is for what controller is used to modify portamento time.

Below that is Key Noise (KN), which works, in a satisfyingly random fashion. Modal resonance was mentioned earlier, and varying it brings in what can best be described as an additional harmonic: You don’t notice it much until you get to higher levels, but when you take it away, you really miss it. Dial it in to taste. Pitch-bend (PB) mode can be asymmetrical (bends down but barely up, like the real instrument), or symmetrical (like on a synth). 

 img
 

Fig. 3. Ms. Sax S, a soprano sax, has similar parameters to the other instruments but is the first to use Samplemodeling’s custom sound engine instead of Kontakt.

The rightmost column starts with sustain pedal assignment, which can control the growl level, sub-harmonic level, or a slightly reduced velocity offset, but Samplemodeling recommends using a continuous controller (aftertouch is nice), not a simple switch. Next is the sub-harmonic level itself, then Touch Sub-harmonic (T.SH), which produces a short sub-harmonic burst at random note attacks, Then, there are the direct levels for growl and flutter, then pitch-bend range. Whew!

Brass are slightly different from saxes, offering control over a variety of mute choices and built-in early reflection impulse controls (see Figure 2). Recently, Samplemodeling has been developing their proprietary SWAM engine (Synchronous Wave Acoustic Modeling), which offers similar controls but, according to the company, is better under the hood for saxophone realism than Kontakt. The first title to offer it is their soprano sax, Ms. Sax S (see Figure 3).


Key-Switches

All the Samplemodeling brass instruments (The Trumpet, The Trombone, and French Horn & Tuba) offer key-switching, which can affect the next note played, the currently sounding note, or what happens at note release. Articulations include accents, crescendos, transient vibrato, short swells, slight vibrato at note release, and fall-offs. The intensity of the effects is usually tied to the velocity of the key-switch trigger itself—for example, the speed and length of the fall-off is tied to how hard you press MIDI note B1. This is a wonderfully expressive way to play. I suggest you add a MIDI pedalboard—such as Keith McMillen’s 12-Step (reviewed Apr.’12)— freeing up your hands. 


Conclusions

In the studio, first you concentrate on playing the right notes, then you go back and add controllers to enhance realism. Live, though, it’s normally very hard to “play” all the requisite controllers in real time. Even using just velocity, pitch-bend, the modulation wheel, expression, and aftertouch, Samplemodeling instruments far outperform most solo instrument libraries I’ve ever tried—making them ideal for live performance.

Is there room for improvement? In some areas, yes. First, I’d like more presets for each instrument. Samplemodeling has given you each instrument in a single, natural presentation, assuming you’ll develop the sound as you play. But I see no harm in rolling say, a dozen presets per title instrument to hook you on the possibilities.

Next, I’d like to see a built-in modulation matrix, as is common in soft synths, so a single controller could be routed to more than one parameter in varying amounts. True, this isn’t needed for “pure” performances, and certainly not when sequencing, but it could add greatly to the live playing experience.

Last, all the instruments have been recorded and voiced with a tasteful film and orchestral leaning. I did get more aggressive tones from them, but I’d love to see some additional sample sets based on stronger jazz, funk, and rock “attitudes.” None of these are failings of the current offerings, just suggestions for a way to expand the palette for the future.

I barely scratched the surface of all that these instruments offer, and I strongly suggest you watch and listen to the many demos on their website. The Samplemodeling instruments are new benchmark of realism and expression for solo wind instruments, and I had a blast playing them all. Note that these are monophonic: You can load multiple instances to create sections, but you can’t just play block chords on a keyboard. So don’t look to them for recording pop-brass and big-band sections in a single pass—there are other libraries for that. But what they do, they do like nothing I’ve ever experienced. These are breakthrough instruments that have raised the bar, and clearly Key Buy winners.




 

You Might Also Like...

Show Comments

These are my comments.

Reader Poll

What kind of vehicle is your gig-mobile?







See results without voting »