A Binaural Reverberator: reverb VST plugins
Author
Girardi, Matteo
Term
4. Term
Education
Publication year
2017
Submitted on
2017-06-02
Pages
60
Abstract
This thesis investigates digital artificial reverberation by developing and comparing two VST plugins built on the Feedback Delay Network (FDN) algorithm. One plugin is a conventional FDN reverberator, while the other spatializes early reflections using a head-related transfer function (HRTF) to add binaural cues aimed at improving the perceived spatial image, particularly for headphone listening. The plugins are prototypes implemented in MATLAB with the Audio System Toolbox, and the report details design choices for early and late components, the mixing matrix and delay lengths, and simple models for air absorption, head shadow, and pinna delays. To assess which approach yields better perceived sound quality, a listening test was conducted and the collected data analyzed (including a repeated-measures analysis of variance). Because this excerpt does not include the results section, no specific findings are reported here; instead, the work frames and tests a practical method for integrating HRTF-based early reflections into FDN reverberation and points toward future low-level implementations and evaluations.
Dette speciale undersøger digital kunstig reverberation ved at udvikle og sammenligne to VST-plugins baseret på Feedback Delay Network (FDN). Det første er en konventionel FDN-reverberator, mens det andet spatialiserer de tidlige refleksioner med en head-related transfer function (HRTF) for at tilføje binaurale spor med henblik på at forbedre den oplevede rumlige lyd, især ved hovedtelefonlytning. Pluginsene er prototyper implementeret i MATLAB med Audio System Toolbox, og rapporten beskriver designvalg for de tidlige og sene dele af reverben, valg af blandingsmatrix og forsinkelseslængder samt simple modeller for luftabsorption, hovedskygge og pinna-forsinkelser. For at vurdere hvilken løsning der giver bedst lydkvalitet er der gennemført en lyttetest, og de indsamlede data er analyseret (bl.a. med variansanalyse med gentagne målinger). Da dette uddrag ikke rummer resultatafsnittet, præsenteres ingen specifikke fund her; arbejdet afgrænser og afprøver en praktisk metode til at indlejre HRTF-baserede tidlige refleksioner i FDN-reverb og peger mod fremtidig implementering i lavniveau-kode og yderligere evaluering.
[This apstract has been generated with the help of AI directly from the project full text]