Introduction
Among string instrument libraries frequently used today, Audio Modeling’s SWAM Solo Strings (SWAM) and Sample Modeling’s Solo, Chamber & Ensemble Strings (SCES) both exhibit a high level of refinement. However, through extended use it has become clear that the underlying philosophies governing how sound is generated and handled are fundamentally different.
At first, I approached them from the common questions of “Which is more realistic?” or “Which sounds better?” But through prolonged use, repeated comparisons, and extensive MIDI programming, it became evident that the two are not in a relationship of superiority or inferiority. Rather, they are instruments built on entirely different assumptions and intended for different roles.
In this article, I would like to organize the differences between SWAM and SCES from the perspectives of the time structure of sound, CC design, and the handling of pp dynamics and long tones.
1. Differences in Design Philosophy
SWAM: A System That Recreates the Act of Performance
WAM is an instrument that calculates physical phenomena such as bow–string friction, energy transfer, and decay in real time.
Parameters such as bow pressure, bow position, bow speed, and vibrato are all directly controlled by the user. Sound is therefore not something “given” in advance, but something that must be continually created moment by moment through performance input.
SCES: A System That Realizes Performance Results as Music
SCES, on the other hand, combines short recorded samples with internal modeling. It can be inferred that it is designed so that its internal state continues to change while a note is sounding.
As a result, by providing cues such as dynamics, phrasing, and vibrato intention, the user triggers sound generation that assumes the ongoing state of a human performer actively playing [Reference ].
2. Why Long Tones “Thin Out” or Do Not
Long tones in SWAM
In SWAM, when CC changes are minimal—that is, when performance conditions tend to become fixed—energy input effectively ceases, and the sound decays naturally. This is a physically correct behavior.
Rather than a flaw, this can be seen as a faithful reproduction of the real-world fact that a tone will thin out unless the performer continues to supply energy through the bow.
Long tones in SCES
In SCES, internal state transitions continue even after the initial attack, maintaining harmonic structure and subtle fluctuations. As a result, the amount of sonic information does not diminish as quickly over time.
Consequently, long tones are less prone to thinning out, and a stronger sense of musical sustain is preserved.
3. Differences in Strength at pp
pp should not be understood simply as “a soft sound,” but rather as a state in which order is maintained with minimal energy.
pp in SWAM
In SWAM, pp involves genuinely low energy and is therefore inherently unstable. If CC design is insufficient, the sound may disappear altogether.
This directly reflects the difficulty of sustaining pp on a real acoustic instrument.
pp in SCES
In SCES, although the volume is low, internal energy is maintained and harmonic content does not collapse.
As a result, pp is preserved as a living, active state rather than a fragile one.
4. Differences in the Meaning of CC Design
SWAM: CC as a Life-Support System
In SWAM, CCs do not primarily function as tools for shaping expression or volume curves. Instead, they act as signals informing the model that the performance is still ongoing.
For this reason, clean straight lines or values that remain fixed for extended periods can actually bring the music to a halt. When CC movement stops, the internal state stabilizes and physical decay begins.
SCES: CC as an Expression of Intent
In SCES, CCs corresponding to bow pressure or bow position do not exist and are presumably optimized internally.
The user only needs to specify how the result should sound. Even if CC movement stops, a minimum level of internal state change continues, allowing musical sustain to be maintained.
In SCES, CCs corresponding to bow pressure or bow position do not exist and are presumably optimized internally.
The user only needs to specify how the result should sound. Even if CC movement stops, a minimum level of internal state change continues, allowing musical sustain to be maintained.
5. Differences in the Treatment of IR and Body
SWAM IR-Body: Reproduces the resonant characteristics of the instrument’s body
SCES Body IR: Cannot be reduced completely to zero; a minimal amount of “musical air” is always present
This is not a limitation, but rather a design philosophy intended to ensure musical viability as an instrument.
6. Practical Usage (Current Conclusion)
At this time, I believe my use of the following.
At present, I use the two systems as follows:
Melodic, singing lines; slow-tempo music → SCES
Inner voices, responses, active passages; faster-tempo music → SWAM
Conclusion
The difference between SWAM and SCES does not lie in whether one is more realistic or more modern, but in a fundamental philosophical distinction:
whether sound is treated as a physical phenomenon or as a musical state.
The issue is not which is superior, but which instrument should be used for which musical role.
Reference
Because SCES can be difficult to understand conceptually, an additional explanation is provided here.
In the main text, I stated that “SCES is designed so that its internal state continues to change while sound is sustained.” Since the internal algorithms are not publicly disclosed, this cannot be asserted at the level of mathematical certainty. However, this interpretation appears to be the most natural, as the following three layers of evidence align.
(1) Design philosophy consistently stated by the manufacturer
Sample Modeling (SCES) consistently emphasizes the following in its official information
Sample Modeling (SCES) has consistently emphasized in its official information that:
- short recorded samples are used,
- long fixed loops are avoided, and
- a hybrid approach is employed in which sound states are reconstructed during performance.
At the very least, it is clear that SCES is not designed to simply replay recorded sounds unchanged along the time axis.
(2) Auditory behavior inconsistent with fixed playback
When listening to long tones, pp passages, or timbral changes under vibrato, SCES sounds are less likely to feel as though they “stop” after a certain duration.
The harmonic structure and internal color appear not to become static. This behavior is difficult to explain solely through looped samples or simple LFO processing.
The impression is that the overtone structure and tonal content are not static. This behavior is difficult to explain by single-sample loop playback or simple LFO processing alone.
(3) The nature of its response to control input
Even with the same note, velocity, and CC settings, the sonic impression may differ depending on performance duration or position within a phrase.
In addition, changes applied mid-note can affect the perceived character of the sound even retrospectively. These observations suggest that the sound is treated not as a fixed result, but as an ongoing, evolving state.
From these points, it seems reasonable to understand that
SCES is designed so that its internal state does not become fixed while sound is sustained, and that—at least perceptually—its timbre and harmonic structure continue to be updated over time.