Real-time physical model of the Rhodes Electric Piano
Author
Sønderbo, Tobias Lykke
Term
4. Term
Education
Publication year
2024
Abstract
This thesis explores the development of a real-time physical model of the Rhodes electric piano using finite-difference time-domain (FDTD) methods and a dynamic grid, combined with MIDI Polyphonic Expression (MPE) to enable per-note control and new layers of expressivity. The model includes core elements such as tine, hammer, and damper; the current pickup implementation is non-physical, with physically based alternatives discussed. A real-time application was implemented with MPE support and optimizations for audio sample rates. Evaluation covers sound and tone production and includes preliminary user interviews. Findings indicate tuning challenges and that the absence of a physically based pickup leads to deviations from the characteristic Rhodes tone, while initial feedback on MPE controls is positive. The work outlines improvements and additions to bring the model closer to the desired output and reflects on trade-offs between expressiveness, fidelity, and computational cost inherent to FDTD.
Dette speciale undersøger udviklingen af en fysisk modelleret Rhodes-elklaver i realtid ved hjælp af finite-difference time-domain (FDTD) metoder og et dynamisk gitter, kombineret med MIDI Polyphonic Expression (MPE) for per-tone styring og udvidet udtrykskraft. Modellen omfatter centrale komponenter som tine, hammer og dæmper; pickuppen er i den nuværende version ikke fysisk baseret, og fysiske alternativer diskuteres. Der er implementeret en realtidsapplikation med MPE-understøttelse og optimeringer til lyd-sample-rater. Evalueringen dækker lyd- og toneproduktion og inkluderer indledende brugerinterviews. Resultaterne peger på udfordringer med stemning og at fraværet af en fysisk baseret pickup medfører afvigelser fra den karakteristiske Rhodes-klang, mens den tidlige feedback på MPE-styringer er positiv. Arbejdet skitserer forbedringer og tilføjelser, der kan bringe modellen tættere på det ønskede output og belyser afvejninger mellem udtrykskraft, klanglig troværdighed og beregningsomkostning i FDTD.
[This apstract has been generated with the help of AI directly from the project full text]