As some major modelers (like Fractal and Line 6) might soon introduce NAM support on their hardware, it could be useful to have metadata for input impedance of the captured device.
This is especially useful for pedal models, since these often have a low input impedance that meaningfully alters the frequency response of the signal, acting as a load on guitar pickups in conjuction with the cable.
Those mentioned modelers, to correctly replicate this interaction between the guitar and the load, when using their own FX blocks, rely on a variable input impedance circuit which is set to the correct value according to the first active block in the chain.
Having metadata for it in the .nam file could allow said modelers to read it and set their input impedance to the nearest value, allowing them to treat NAM models like their proprietary fx blocks in this regard.
Some might argue that .nam models of devices with low input impedance already contain such alterations of the frequency response when captured with a reamp box, but this is a simplistic and incorrect way to solve the problem:
-
Because a reamp box, even if it has an output impedance comparable to that of a guitar, can't be the same as any guitar a user could plug in an amp or pedal (a guitar pickup can have an impedance as low as 3-4k and as high as 20k, or maybe more, there's too much variance)
-
Because, even if it has exactly the same output impedance as a specific guitar pickup, it won't produce the same frequency response alterations as a guitar pickup, cuz it can't have the same resonant peak shifting and dampening which are produced by the inductance of the pickup.
-
Any coloration embedded in the NAM model would be added on top of the already more or less coloured frequency response of a guitar connected to the instrument input of an audio interface or modeler, with the risk of creating excessive and redundant filtering.
So, the best way to create models is to use a low impedance output while reamping the test signal and then eventually lower the input impedance of the device used to play the model.
Exactly the same approach commonly used for traditional "white box" nodeling.
As some major modelers (like Fractal and Line 6) might soon introduce NAM support on their hardware, it could be useful to have metadata for input impedance of the captured device.
This is especially useful for pedal models, since these often have a low input impedance that meaningfully alters the frequency response of the signal, acting as a load on guitar pickups in conjuction with the cable.
Those mentioned modelers, to correctly replicate this interaction between the guitar and the load, when using their own FX blocks, rely on a variable input impedance circuit which is set to the correct value according to the first active block in the chain.
Having metadata for it in the .nam file could allow said modelers to read it and set their input impedance to the nearest value, allowing them to treat NAM models like their proprietary fx blocks in this regard.
Some might argue that .nam models of devices with low input impedance already contain such alterations of the frequency response when captured with a reamp box, but this is a simplistic and incorrect way to solve the problem:
Because a reamp box, even if it has an output impedance comparable to that of a guitar, can't be the same as any guitar a user could plug in an amp or pedal (a guitar pickup can have an impedance as low as 3-4k and as high as 20k, or maybe more, there's too much variance)
Because, even if it has exactly the same output impedance as a specific guitar pickup, it won't produce the same frequency response alterations as a guitar pickup, cuz it can't have the same resonant peak shifting and dampening which are produced by the inductance of the pickup.
Any coloration embedded in the NAM model would be added on top of the already more or less coloured frequency response of a guitar connected to the instrument input of an audio interface or modeler, with the risk of creating excessive and redundant filtering.
So, the best way to create models is to use a low impedance output while reamping the test signal and then eventually lower the input impedance of the device used to play the model.
Exactly the same approach commonly used for traditional "white box" nodeling.