This is a general introduction about analog modling, its advantages and concepts that one should know when using it.
What is analog modeling?
Analog modeling is a method to emulate the tone of an analog instrument or sound processor by means of a digital system, like a computer or a
DSP. With analog modeling it is possible to reproduce the same tone characteristics and the same parametric behaviour of the modeled analog
instrument, doing so on a computer or on a digital processor. In this way we can reach an important advantage, by joining the best of the analog and
digital worlds: the sound quality and euphonic character inherent of the analog processors plus the versatility and processing capabilities of digital
systems. Examples are total recall of setups and the ability to load of as many instances of the effect as possible.
The real analog modeling, the one we’re proud to implement, starts from the electrical schematic of the instrument to be simulated: each component of
the schematic has its own mathematical model. Then by appropriate methods these are translated into a system of equations and finally into a
computer or DSP program. After this rather raw scientific step, we fine tune the resulting tone by comparing it to the original instrument with real life
audio samples.
We’d like to point out that, even if the majority of developers claim to deliver analog modeling, only a few are really serious about it: in many cases
developers use standard widely available DSP techniques and obtain a tone which only vaguely resembles the original one, but it is actually different.
For example, the filters are cutting at the same frequencies as the hardware unit, but their shapes and resonances can be rather different. On the other
hand, Overloud is using its own developed non-standard DSP techniques [1], [2], [3] which allows us to reproduce every detail of the original tone.
Analog Modeling advantages
Analog Modeling joins the best of both worlds: tone quality directly comparable to the modeled instrument and the widely known versatility of digital
platforms. The main pluses are:
- faithful reproduction of every detail of the analog tone;
- perfect reproduction in parameters behaviour: user’s aural feedback is exactly the same as if she/he’s working on the real instrument;
- you can easily save and recall the knob position and manage large amounts of presets;
- you can load as many instances as the host processor/DSP allows, without the need to invest in more than one hardware to process multiple tracks
at the same time;
- it is much less expensive than a hardware unit;
- you can quickly add more features or high quality effects to a software or to a DSP; this task only requires a change in the software program.
What you should be aware of...
We do not want to state that analog modeling is the definitive solution to replace all kinds of hardware.
This is a more technical section about things one should be aware about analog modeling.
A first, general and simple rule is the following one: analog modelling is able to handle perfectly all the processes where both the input and the output
are electrical signals. For example analog compressors, equalizers, stomp-boxes and similar devices accept an incoming sound with an electrical
connection and produce an electrical output. Likewise, synthesizers carry an electrical input (the MIDI in or the keyboard) and an electrical output. In all
these cases, the electrical signal can be represented digitally and can be processed by the DSP.
On the other hand, consider the following three cases:
1) Microphone simulation: in this case you have an acoustic input (the sound pressure wave) and an
analog output (the electrical output of the microphone). If you record an instrument with an SM58, which is band-limited to 50-15000Hz, you won't be
able to simulate a U87 which has a frequency response of 20-20000Hz, because the harmonics in the 15-20KHz are missing in the original recording.
Analog modelling cannot recreate a piece of information which is lost in the transducer. Here, the weak element in the chain is the transducer, which is
not able to handle all the sound content that the digital processing can manage.
2) Guitar amplifier simulation: in this case you have an electrical input (the voltage at the guitar
pickups) and an acoustic output (the sound pressure in front of the speaker cabinet). Suppose you are playing a simulation of a high gain amplifier with
a 4x12 cabinet speaker and you are listening it through multimedia speakers or even on the laptop speakers; you won’t be able to experience the same sound impression that you have standing in front of a real 4x12 cabinet. Also in this case, the weak part is the transducer (the speaker) which is not able to handle all the sound content generated by the digital processing. So, how are guitar amplifier simulation software usually working? All the guitar amp simulators are based on a sound chain which comprises the amplifier, the speaker cabinet and a microphone which picks-up the sound in a certain position in front of the cabinet. With this configuration, the input is the electrical signal coming from the pickups of the guitar. The output is also an electrical signal, the one at the output of the microphone. What you always obtain from a guitar amp simulator is the tone of the microphoned amplifier, like the one you have when record a guitar in a studio.
3) Microphone preamplifier simulation: different models of microphone preamplifiers usually present quite different input impedances. Some preamplifiers let the user select even the desired input impedance. These differences directly affect the physics of the microphone, which is loaded with a variable impedance. When you are using a simulated microphone preamplifier, the microphone is physically connected to an unknown preamplifier (it can be the input on the mixer or an outboard preamplifier, or something different), and of course it makes no sense to connect the microphone directly to the model, because it is a software component. So it is not possible to physically interact with your microphone like the real unit does. One of the known turnarounds for this issue is to code into the algorithm the behaviour of different type of microphones connected to that particular model. In this way you do not have a generic model of the “model x” preamplifier, but you have several models, one for “model x” connected to a SM58, another for “model x” connected to a U47, and so on… The disadvantage is that it is almost impossible to cover all the existing microphones, and one should stop after only the most common ones are emulated.
Bibliography
[1] T. Serafini, "A complete model of a tube amplifier stage", Simulanalog Technical Report
[2] T. Serafini, P. Zamboni, "State variable changes to avoid non computational issues", Simulanalog Technical Report
[3] S. Barbati, T. Serafini, "A perceptual approach on clipping and saturation", Simulanalog Technical Report |
