Adaptive Modulation in a Digital Modulation Plugin: Design, Implementation, and Evaluation
Author
Einarsson, Thorvald Johan
Term
4. Term
Education
Publication year
2026
Submitted on
2026-05-26
Abstract
Modality Machine is a real-time guitar audio effect, implemented as a VST3 plugin in C++ using the JUCE framework. The thesis describes the system's design, iterative implementation, and evaluation. The central research question is whether a real-time detector of single notes and chords can serve as a musically meaningful control signal in a guitar effects context, and whether the resulting adaptive modulation feels expressive and perceptually coherent to the performer. The main technical contribution is a hybrid detector that distinguishes, in real time, between melodic single-note playing and harmonic chord playing. It combines measures such as spectral flatness and a log-frequency band-density metric with a chord-hold mechanism using hysteresis. Put simply, the detector analyzes the distribution and stability of the sound to infer what is being played. The detector outputs a continuous complexity signal that is used as a general musical control source. In the presented mapping, this signal controls stereo width: single-note lines remain spatially focused, while chords expand into a wider stereo field. Another adaptive mapping links the amplitude envelope to modulation depth, so the intensity naturally follows the player's dynamics. The plugin offers two selectable modulation effects—chorus and tremolo/panner—and a two-part user interface. A performance view provides key controls, while a calibration mode exposes the detector's internal state in real time so the user can adjust the system. The work sits within adaptive digital audio effects and new interfaces for musical expression, contributing a new detection architecture, a principle-based interaction design framework for adaptive guitar effects, and a structured methodology for evaluating performer responsiveness in sound systems.
Modality Machine er et realtids-lydplugin til guitar, udviklet som et VST3-plugin i C++ med JUCE. Specialet beskriver design, iterativ implementering og evaluering af systemet. Kernen i arbejdet er et forskningsspørgsmål: Kan en realtidsdetektor for enkelttoner og akkorder bruges som et musikalsk meningsfuldt kontrolsignal i en guitareffekt, og opleves den adaptive modulationsadfærd som ekspressiv og perceptuelt sammenhængende for performeren? Det primære tekniske bidrag er en hybrid detektor, der i realtid skelner mellem melodisk spil med enkelttoner og harmonisk akkordspil. Den kombinerer mål som spektral fladhed og et log-frekvensbaseret båndtæthedsmål med en akkord-holdemekanisme med hysterese. Forenklet sagt analyserer detektoren klangens fordeling og stabilitet for at vurdere, hvad der spilles. Detektoren producerer et kontinuerligt kompleksitetssignal, som bruges som generelt musikalsk kontrolsignal. I den konkrete realisering styrer signalet stereobredden: enkelttonelinjer forbliver rumligt fokuserede, mens akkorder udvides til et bredere stereofelt. En anden adaptiv mapping kobler amplitudekuverten til modulationsdybden, så modulationsintensiteten naturligt reagerer på spillerens dynamik. Pluginet rummer to valgbare modulationseffekter, chorus og tremolo/panner, samt en todelt brugergrænseflade. Performancevisningen samler centrale kontroller, mens en kalibreringstilstand viser detektorens interne tilstand i realtid, så brugeren kan justere systemet. Arbejdet er forankret i felterne adaptive digitale lydeffekter og nye grænseflader for musikalsk udtryk. Det bidrager med en ny detektionsarkitektur, et principbaseret interaktionsdesignframework for adaptive guitareffekter og en struktureret metodik til at evaluere performer-responsivitet i lydsystemer.
[This abstract has been rewritten with the help of AI based on the project's original abstract]
Keywords