Back To Schedule

Spice Up Your Audio

Accurate Component Modeling for Realtime Effects

What if you could design an analog circuit and then deploy it with the ease of software, without constructing a physical embodiment? This talk will equip audio developers with the knowledge and practical steps to leverage SPICE models for creating highly accurate and unique realtime audio effects.

While digital audio effects often rely on idealized component models, this can lead to a sterile, clinical and unrealistic sound. Real-world components have imperfections and non-linearities that contribute to their musical character. SPICE (Simulation Program with Integrated Circuit Emphasis) allows us to model components with incredible accuracy, capturing these nuances.

We will cover the entire process: from sourcing data to building a realtime audio plugin. Attendees will learn to model vintage or rare components accurately,
understand the impact of component tolerances on sound, and leave with a functional workflow for integrating SPICE into realtime audio processing. Along the way, we will attempt to keep technical information to the absolutely necessary, ensuring the knowledge is accessible to all levels of developers.

We begin with a short explanation of SPICE and cover the various SPICE engines. Then we explain the tradeoffs between behavioral and component-level models, with appropriate audio examples. We also show where to sourcing model data, such as datasheets, books, app notes, and papers. Of course, the gold standard is hardware measurements. When in doubt, measure!

In addition, we will showcase how to derive and validate your SPICE models, with examples such as transistors. An important consideration is the precision of the modeling calculation as a whole - not to be confused with the precision of individual operations within. As such, empirical hints will be given on how to set the tolerances of your simulation.

The last part will examine how to convert the simulation to a realtime plugin. We will look at computational efficiency, approaches that leverage pre-computation, and creating simplified models for audio. Next, we will implement our SPICE model in code, showing how to integrate it into a larger project.

Finally, provided there is enough time, there will be a realtime audio plugin demo featuring component-level SPICE simulation.