In an earlier announcement regarding the update of the physical-modeling string instrument SWAM to version 3.10, the manufacturer stated that IR (Impulse Response) had been introduced. However, when I checked the official release notes, I noticed that the term IR does not appear there at all; instead, the term Body is used.
This discrepancy caught my attention, so I revisited the available manufacturer information and tried to organize my own understanding of the situation.
What follows is a revised version of Part 2.
If there are any inaccuracies in my interpretation, I would be grateful for your corrections.
1. SWAM’s Design Philosophy and the Previous System (up to v3.9)
SWAM is a fully physical-modeling instrument that generates sound through real-time computation by modeling the actual structure of the instrument and the performer’s playing actions.
Based on inputs such as bow movement and pressure, finger position, and articulation,
the resonance of the strings and the body, as well as changes in harmonic content, are continuously calculated.
As a result—although the differences may be subtle and not always easy to identify— even the same MIDI data can produce slightly different results each time. While CPU usage is relatively high, the library size itself is extremely small.
Earlier versions of SWAM included an Ambience section (internal room simulation), with presets named after cities such as Cremona, Firenze, and Roma, allowing users to change the sense of space and distance.
This Ambience was not based on convolution IR,
but rather on an algorithm-based spatial model.
When Ambience was turned off (or its Amount set to the minimum),
the spatial reverberation disappeared, leaving only the sound generated by the physical modeling engine.
In such cases, it was possible to apply external IRs or reverbs on the DAW side if desired.
In version 3.8 (2024), the Ambience (Room Simulator) was redesigned, making more advanced room simulation possible; however, at that stage IR was still not being used.
2. What Changed in SWAM v3.10: The Relationship Between Body and IR
In SWAM v3.10, released in 2025,
the official release notes do not use the term IR,
but they clearly state that new Bodies have been added.
At the same time, recent promotional material includes the following description:
Solo Strings now include brand-new Impulse Responses, simulating the acoustic body of each instrument with remarkable realism.
From this, it can be understood that
new IRs simulating the acoustic characteristics of the instrument body have been introduced for each Solo String instrument.
That said, these IRs are not provided as a standalone feature that users can directly manipulate or replace.
Instead, Audio Modeling appears to have integrated them into a musical and conceptual unit called Body,
which is presented in the user interface as city-named presets.
The Dry Body mentioned in the release notes can be interpreted as a neutral reference state, in which these acoustic characteristics (the Body component including IR) are not applied.
Addition:
After contacting Audio Modeling to clarify the difference between the previously existing Bodies and the newly added ones, the following points became clear.
In SWAM v3.10, new Bodies based on measured impulse responses have been introduced.
On the other hand, previously existing Bodies such as Cremona have remained internally unchanged in order to ensure full backward compatibility with older projects.
Audio Modeling refers to both of these as “IR-based Bodies,” but it is important to note that there are differences in the underlying implementation generations.
At the same time, SWAM provides separate Sound presets (such as Bach or Saint-Saëns) that define how the instrument is played,
after which the Body—selected by city name—determines how the instrument itself sounds.
In v3.10, IRs representing near-field acoustic characteristics centered on the instrument body were introduced into the internal implementation of these Bodies,
while the overall operation and workflow remain unchanged from earlier versions.
Accordingly, the signal structure in v3.10 can be understood as a multi-layered system:
– Physical modeling sound generation (playing style and performance behavior)
- Body (an IR-based character including acoustic properties of the instrument body)
– Ambience (room simulation handling space and distance)
- Output
Because the release notes avoid the implementation term IR and instead use the user-oriented concept of Body, a mismatch in wording may have occurred between different forms of official communication.
3. The Development Paths of Sample Modeling and Audio Modeling
At this point, it may be useful to briefly look back at the history.
3-1. The Emergence of Sample Modeling
Around 2007, Sample Modeling was founded by a group of Italian engineers.
They adopted a hybrid approach that combined very short samples with physical models for correction and transformation.
Wind instrument sound sources such as The Trumpet and The Saxophone have received high praise for achieving both realistic samples and flexible performance expression.
3-2. Independence of Audio Modeling
Around 2017, some of the co-founders of Sample Modeling left to establish a new company, Audio Modeling. It was here that the current SWAM engine was developed.
Audio Modeling pursued the direction of complete physical modeling that uses no samples whatsoever.
As a result,
– Sample Modeling: Sample + Physical Model (Hybrid)
– Audio Modeling (SWAM): Full Physical Modeling
Two distinct trends emerged.
4. Differences in the Use of IR
This divergence is also reflected in how IR is handled.
Sample Modeling (SCES)
– Based on samples recorded in an anechoic chamber
– Adds spatial characteristics via internal IR (Convolutional Reverb)
– The IR contribution can be controlled via CC100
Audio Modeling (SWAM)
– Sound generation itself is fully physical modeling
– In v3.10, IRs representing near-field acoustic characteristics centered on the instrument body were integrated internally as Bodies
– IR is not an independent function but integrated as part of the tone character
It is fascinating that despite sharing the same roots, SCES arrived at "samples + spatial IR," while SWAM reached "physical sound generation + instrument body IR + spatial simulation."
5. Future Initiatives
Previously, I tried inputting Saint-Saëns' "Le Cygne" using SWAM Solo Cello. This time, I want to see how the Body feature (IR-based acoustic characteristics) introduced in v3.10 affects this piece by comparing it with the Dry Body. However, my input skills remain amateurish, so I plan to focus solely on examining the differences in sound.
Note
This article is based on the manufacturer's publicly available information (Release Notes and promotional materials) and actual usage experience. If you notice any technical inaccuracies or have additional information, please let us know.