SGT-DI: Ampeg SGT-DI NAM captures
Download the free NAM A2 pack of the Ampeg SGT-DI bass preamp or the RedGrooveBass plugin, which already has it inside. Or just listen below and download the captures one by one.
The device
Ampeg SGT-DI
| Serial number | Year | Origin |
|---|---|---|
| 21SGD2V7522000062 | 2025 | Designed in the United States, made in China |
- Captured by
- Darwin's Cat
- Captures
- 22
- Plays in
- OrbitBass, RedGrooveBass
Ampeg SGT-DI
- Character
- on the edge of breakup · overdrive
| Grit | Voice | Epochs | ESR | Download |
|---|---|---|---|---|
| 7h | SVT | 800 | 0.00109 | Ampeg SGT-DI 7h SVT.nam |
| 8h | SVT | 800 | 0.000913 | Ampeg SGT-DI 8h SVT.nam |
| 9h | SVT | 800 | 0.00122 | Ampeg SGT-DI 9h SVT.nam |
| 10h | SVT | 800 | 0.00127 | Ampeg SGT-DI 10h SVT.nam |
| 11h | SVT | 800 | 0.0025 | Ampeg SGT-DI 11h SVT.nam |
| 12h | SVT | 800 | 0.00247 | Ampeg SGT-DI 12h SVT.nam |
| 13h | SVT | 1000 | 0.00449 | Ampeg SGT-DI 13h SVT.nam |
| 14h | SVT | 1000 | 0.00773 | Ampeg SGT-DI 14h SVT.nam |
| 15h | SVT | 1200 | 0.0104 | Ampeg SGT-DI 15h SVT.nam |
| 16h | SVT | 1200 | 0.012 | Ampeg SGT-DI 16h SVT.nam |
| 17h | SVT | 1200 | 0.0122 | Ampeg SGT-DI 17h SVT.nam |
| 7h | B15 | 800 | 0.000534 | Ampeg SGT-DI 7h B15.nam |
| 8h | B15 | 800 | 0.000528 | Ampeg SGT-DI 8h B15.nam |
| 9h | B15 | 800 | 0.000665 | Ampeg SGT-DI 9h B15.nam |
| 10h | B15 | 800 | 0.000944 | Ampeg SGT-DI 10h B15.nam |
| 11h | B15 | 800 | 0.0013 | Ampeg SGT-DI 11h B15.nam |
| 12h | B15 | 800 | 0.00207 | Ampeg SGT-DI 12h B15.nam |
| 13h | B15 | 800 | 0.00259 | Ampeg SGT-DI 13h B15.nam |
| 14h | B15 | 1000 | 0.00475 | Ampeg SGT-DI 14h B15.nam |
| 15h | B15 | 1000 | 0.00583 | Ampeg SGT-DI 15h B15.nam |
| 16h | B15 | 1000 | 0.00728 | Ampeg SGT-DI 16h B15.nam |
| 17h | B15 | 1000 | 0.00774 | Ampeg SGT-DI 17h B15.nam |
Tone stack
- minimum
- reference
- maximum
- minimum
- reference
- maximum
- minimum
- reference
- maximum
- -
- 0
- +
- 0
- +
Sky in the picture: NASA/Goddard Space Flight Center Scientific Visualization Studio; Gaia DR2: ESA/Gaia/DPAC
Frequently Asked Questions
Where do "Great!", "Not bad!" and the other ESR ratings come from?
ESR (error-to-signal ratio) is how far the model strays from the recording it learned from: the energy of the difference divided by the energy of the sound itself, on a validation slice of the recording. The smaller, the closer.
| ESR | rating |
|---|---|
| below 0.01 | Great! |
| below 0.035 | Not bad! |
| below 0.1 | ...This *might* sound ok! |
| below 0.3 | ...This probably won't sound great :( |
neural-amp-modeler · nam/train/core.py
An ESR of 0.01 means the difference is 20 dB quieter than the sound itself; 0.005 means 23 dB.
What is a training curve?
How the model's ESR changed while it learned. Across: the epochs, each one a pass of the trainer over the whole recording. Up: the ESR on the validation slice after each epoch, on a logarithmic scale.
At first the ESR falls fast, then the curve flattens out: the model is learning the fine details. The spikes are single epochs where the training briefly went off track; after them the curve comes back to where it was.
A capture's ESR is the number the trainer gave for the finished model. Hover over the curve to see the ESR at every epoch.