Physical Models¶
Three struck physical models ship with MAGDA: Marimba, Djembe, and Bell. Each is a modal physical model driven by a strike exciter, not a sample. The ring and decay come out of the model itself, so there are no loops or release samples.
All three are 16-voice instruments. Marimba and Djembe follow the played note for pitch. Bell is fixed-pitch: the note only triggers the strike.
Marimba¶
A struck tone-bar-and-tube model.
| Parameter | Range | Default | Description |
|---|---|---|---|
| Strike Position | 0 to 1 | 0.3 | Strike position on the bar (0 centre, 1 edge) |
| Strike Tone | 500 to 12000 Hz (log) | 7000 | Exciter brightness |
| Strike Sharpness | 0 to 1 | 0.25 | Exciter transient hardness |
| Decay | 50 to 2000 ms | 100 | Bar ring time (T60) |
Djembe¶
A struck hand-drum membrane model.
| Parameter | Range | Default | Description |
|---|---|---|---|
| Strike Position | 0 to 1 | 0.4 | Strike location (0 centre, 1 edge) |
| Strike Sharpness | 0 to 1 | 0.5 | Exciter transient hardness |
| Decay | 50 to 3000 ms | 600 | Lowest-mode ring time (T60) |
| Spacing | 20 to 600 Hz | 200 | Modal frequency spacing (higher is more metallic) |
| Inharmonicity | 0 to 1 | 0 | Upper-mode spreading (0 even, 1 widest) |
At low Spacing the Djembe reads as a pitched tom; raise it for a more metallic, gong-like character.
Bell¶
A struck, fixed-pitch modal church-bell model.
| Parameter | Range | Default | Description |
|---|---|---|---|
| Strike Position | 0 to 1 | 0.3 | Strike position on the bell (0 centre, 1 edge) |
| Strike Tone | 500 to 12000 Hz (log) | 7000 | Exciter brightness |
| Strike Sharpness | 0 to 1 | 0.25 | Exciter transient hardness |
| Decay | 300 to 40000 ms | 8000 | Ring time (T60); cathedral-long at the top, tine-short at the bottom |
Macros, modulation, and presets¶
Each model is a standard MAGDA device, so its parameters accept macros and modulators, and patches save and recall through the preset system.
AI sound design¶
Click the AI icon in the device header and describe the sound you want. The agent reads this device's own parameters and their real-world ranges, then writes values into them. See Per-Device AI Panel.