Modelling the perception of percussive low frequency instruments in rooms

Throughout the study of room acoustics, adverse effects and methods of controllingand mitigating modal behaviour are a well researched topic. However, despite this, agap between objective metrics and subjective results is still prevalent, thus resultingin a limited understanding of perceived bass quality.Previous work has suggested a group of perceptual attributes that are useful indescribing the effect of room acoustics on the perceived bass quality, however anobjective link to the perceptual attributes has not been quantified. Furthermore,the scope of previous work is mostly concerned with small listening rooms and rarelyextends to other cases, such as that found in live sound reinforcement.Hence, this work is focused on broadening the understanding of low frequencyquality due to modal behaviour in rooms, through extending the scope of research toinclude larger listening environments and single instrument excitation of the room.To investigate the characteristics of low frequency quality, various kick drumswere auralised using an improvement to the modal decomposition model and werethen rated in a subjective listening test using the descriptive bass quality attributes.From the results, the attributes were modelled through a novel approach using aRandom Forest model, utilising a combination of acoustic and MIR features.It was found that the perceptual attributes of both Resonance and Articulationwere predicted effectively from signal features, however Bass Energy was unable tobe modelled with any accuracy. Use of feature selection algorithms revealed thatResonance and Articulation attributes relied on temporal and decay based features,such as early decay time and temporal centroid. This result further suggests theimportance of temporal modal behaviour when considering audible effects due tolow frequency modes. The outcome of this work supports the growing body of workthat the effects of modal density are not as important as traditionally thought andis therefore applicable to both small and large rooms.

Level dc:type.qualificationlevel

Master's Level (Level 7)

Year dc:date.issued

2020

Author dc:creator

• Howard, MHR

Language dc:language

en

Identifier

oai:salford-repository.worktribe.com:1347011

Author Identifier

0000-0001-6983-6322

OAI identifier oai:identifier

oai:salford-repository.worktribe.com:1347011