An ongoing trend for sample libraries throughout the years is that they are always getting larger! Musicians demand better quality samples, more articulations, finer control, and overall more realism in their virtual instruments. The two prominent methods of synthesis for instruments are wavetable-based libraries and physically modeled ones. While the first option offers more attainable realism, the latter ultimately has more flexibility.
Sample libraries have become more extensive, including more and more articulations, usually accessible by keyswitches. New technology has emerged to improve realism, such as release samples and round-robin alternating samples. Physical modeling has become more elaborate, reproducing more detailed sound models and accounting for more aspects of physics. However, both sampled instruments and physically modeled instruments face the same dilemma: the performer itself must be modeled.
The first problem to face is what to do with limited MIDI data, from the sequencer or MIDI keyboard. MIDI is a very keyboard-oriented format, which makes it naturally limited in terms of controls. Basically, you can control the pitch of the note, velocity, as well as a few controllers (including the pitch wheel). However, you are limited as to how many controllers you can use simultaneously in a live performance.
Example: Iconic Bass: Jaco uses the sustain pedal to activate the slide mode, which changes any overlapping notes into a smooth glissando. The speed of the slide is based on the velocity of the overlapping note. The ability to slide from one note to another is a important characteristic of the fretless bass.
Due to the limitations of simultaneously accessible MIDI controllers, some automatic interpretation needs to be made by the sample library. For example, which frets and strings to use in a guitar library must be intelligently determined. Bowed instruments must be able to control bow pressure, speed, and direction. Although these controls can be assigned to MIDI controllers, the practicality of having all these elements instantly controllable must be questioned, as to not limit the use of sample libraries to sequencing but include live performance from a MIDI keyboard. While sequencing allows you to fine tune every element of the sample library dynamically, having too many controls to keep track of can be overwhelming and counterproductive.
Example: Strawberry Electric Guitar includes a layer-based mapping system which allows you to define exactly how the instrument behaves. It includes controls for how to translate MIDI data as well as letting you adjust the automatic interpretation.
Another thing to consider is the limitations of not only the instrument, but the performer as well. Most sample libraries account for pitch range limitations, but how many model hand movement speed and the maximum reach of the player’s hand? These both are important considerations in most instruments.
Modeling player limitations is essential when interpreting MIDI to automatically control elements of the instrument’s performance. That way the scripting engine for the sample library has a framework to anticipate and automate characteristics of the instrument beyond the instant control of the MIDI keyboard or MIDI sequencing. This can involve controlling something as obvious as articulations…
Example: CoreBass Cherry Slapped models the thumb and index finger position in order to accurately determine whether a note should be slapped or popped in true slap bass playing style.
…or details such as fret noises:
Example: CoreBass Pear tracks the hand position on the fingerboard, not only to correctly decide which strings to use when playing, but to model the finger position to reproduce the sliding noise created when switching positions. The scripted finger positions follow traditional upright bass playing technique.
Modeling the player in virtual instruments allows users who aren’t familiar with how the particular instrument is played to create realistic tracks. It also overcomes the characteristic artificial sound that virtual instruments tend to have, by replicating essential humanistic elements that would otherwise be too difficult and time-consuming to sequence manually.
Tags: CoreBass Cherry Slapped, CoreBass Pear, iconic bass jaco, realism, sampling, Strawberry Electric Guitar
Greg,
Very good read and spot on. I think the solution is midi controllers that act more like the real instrument… for example, I feel it is an exercise in futility to try and create a realistic guitar part using a keyboard and controllers. Yes, some library contain “contrived” artics that can be manipulated with controller data or keyswitches but in all cases, the artic is limited to the “wave” file recordings timing and feel.
I am a big fan of MIDI guitar controllers to try and trigger guitar samples more realistically… it does a great job in quick slides and note lengths and timings but alas, it converts note bends or long slides to a series of pitch bend data (which is far more realistic than trying to manually do a note bend using a pitch wheel). But even this leave alot to be desired since you are tied to the sounds of the sample set only.
So, some type of sampled/physical modeled instruments are part of the answer but the other part is more intelligent midi controllers and for that matter, a new MIDI controller standard to fit the player styles of the new controllers as well.
This is just my opinion and I hope that you can help further this technology to create more realistic ways to create “virtual” instrument tracks. I know that the argument can be made to why not just play a real guitar and record it and it is a valid argument. My counter is, how can anyone afford 20 or 30 different high quality guitars in their tool box when each one costs thousands of dollars and some as much as $5K????? At least for me, not very affordable and my 7 guitar arsenal has good me a fortune and I don’t feel like laying out another $20K for more guitars. But imagine a controller that would interpret your playing perfectly and then be able to model any of these guitar virtually? Amazing stuff and I can’t wait.
Jim
Hi Jim,
I agree! I think the MIDI guitar world has a lot of possibilities to be explored. I think there is a potential market for sampled (or physically modeled) guitar libraries for guitarists, for example.
Imagine being able to load up a vintage Les Paul library and switch to a Martin acoustic guitar library with the press of a foot switch. These virtual instruments would give you convenient access to guitars you would might otherwise not be able to afford. Of course, it wouldn’t emulate the proper feel of the guitars while you’re playing them, but considering the price difference, it could be a cost-effective alternative for studio and performing guitarists alike.
Best regards,
Greg
Good article. I must say I’m very impressed with what you’ve done with Iconic Bass: Jaco to more closely emulate a real bassist, bravo! It would certainly fool me with ease on any track. I think to a great degree modeling player performance is not only limited by the way different sampled articulations are triggered but the method used to trigger them. The MIDI world is dominated primarily by keyboard and drumset based instruments. Just as wind controllers add a level of realism to horns and woodwinds, I think the world of plucked string based MIDI instruments needs to be further explored in order to capture many of the performance nuances of those instruments. Then again, if I was a Kontakt user I wouldn’t hesitate to pick up one of your fantastic bass libraries. Again, great job!
Will, thanks for the compliments–you’ve made my day!
Hey there, great article… You’ve definitely addressed something I’ve thought about many many times but could never think of. Have you seen anything about OSC (Open Sound Control)? It seems to trump MIDI quite a bit and provide a much more expressive set of possibilities…. I think it’s the future.
Sounds cool–I’m reading about Open Sound Control right now.
I have recently been dabbling in sound and this is a great site!