A framework to support the development, in an educational environment, of real-time, digital sound synthesis algorithms is proposed. Sound synthesis algorithms are an important subset of Digital Signal Processing (DSP), and are an excellent way of teaching the application issues of many DSP concepts. Steinberg’s Virtual Studio Technology (VST) is a very flexible format for creating digital sound synthesis and audio effect plugin applications. The company provides an associated C/C++ SDK, and an open source wrapper (jVSTwRapper) is available that allows the plugin code to be written in Java. However, the jVSTwRapper documentation is sparse, and examples bundled are difficult to extrapolate from, reducing its effectiveness in an educational context. A simpler architecture allows synthesis algorithms to implemented more easily, keeping the focus on the algorithms themselves and not on customising the supporting VST code. This paper proposes such an architecture, comprising a novel generalised voicing structure for the JVSTwRapper, taking advantage of the benefits that the Java language offers for students.