Listening Examples: Gray-Box Modeling of Guitar Amplifiers

In this work, analog guitar amplifiers were modeled with an automated procedure using input/output measurements and iterative optimization. The used digital model is called a Wiener–Hammerstein model consisting of a linear time-invariant (LTI) block, a nonlinear block with a nonlinear mapping function and another LTI block connected in series. The model is adapted in two steps. First the filters are measured and afterwards the parameters for the nonlinear part of the digital model are optimized with the Levenberg–Marquardt method to minimize a cost-function describing the error between digital model and analog reference system. A small number of guitar amplifiers was modeled in different settings. The tone section of each amplifier was set to a 12 o’Clock position and gain and volume controls were altered from a low value (9 o’Clock position) to medium value (12 o’Clock position) to high value (3 o’Clock position).

Different guitars using different pick-ups were used to create the listening examples. Some self-recorded input signals were used and some input signals were taken from the Fraunhofer IDMT guitar database. The used guitars were an Ernie Ball stratocaster-style guitar with singlecoil pick-ups (SC), a Reverend — Buckshot with humbucker pick-ups (HM1), a Hagstrom — Super Swede with Gibson humbuckers (HM2) featuring a higher output voltage than the reverend pick-ups and finally the Ibanez — RG2820 with DiMarzio humbuckers (HM3) from the IDMT-SMT guitar dataset.

Short listening examples are presented below:

Roost SR22 (Crunch)

High Gain, Medium Volume – HM2

Reference:

Model:

 

High Gain, Medium Volume – HM3

Reference:

Model:

Madamp A15Mk2

Low Gain, Low Volume – HM3

Reference:

Model:

 

Low Gain, Low Volume – Bass

Reference:

Model:

 

Medium Gain, Medium Volume – HM3

Reference:

Model:

 

Medium Gain, High Volume – HM3

Reference:

Model:

 

High Gain, High Volume (+ Gain Boost) – HM3

Reference:

Model: