A Review of String Instrument Synthesis Methods for Use in Interactive Systems

This article presents a scoping review of acoustic string instrument sound synthesis for use in interactive systems, covering studies from 1983 to 2025. Following the PRISMA-ScR checklist, we identified 67 eligible works and categorized them into three tiers based on their implementation status with respect to interactive use. To compare the synthesis methods, we introduce a four-dimensional evaluation framework capturing fidelity, responsiveness, controllability, and adaptability. Results show that physical modeling dominates plucked string sound synthesis, as the excitation–resonator interactions in plucked string instruments can be compactly captured by parametric representations. In contrast, neural audio synthesis proves more effective for bowed and hammered string instruments due to the highly nonlinear, time-varying dynamics of bowed-string excitation and the complex transient and resonant behavior of hammered-string instruments, both of which benefit from data-driven modeling. Today, abstract digital methods appear less frequently, reflecting their limited capacity to reproduce fine-grained nuances of instrument sounds. As a central challenge, documentation of latency and responsiveness remains inconsistent in the literature, hindering reproducibility and comparative evaluation across systems. Overall, our review highlights the tradeoffs between fidelity, responsiveness, controllability, and adaptability and argues that clearer reporting of implementation constraints, more consistent evaluation protocols, and explicit consideration of cross-instrument generalization offer promising paths for advancing interactive string instrument sound synthesis.