Marc Ciufo Green

Psychoacoustic metrics offer the means to analyse sound qualities using sophisticated models of human perception. A substantial drawback is the need for high-resolution acoustic data as input. Quasi-psychoacoustic metrics are proposed to address sound quality analysis when high-resolution acoustic data are unavailable. The proposed metrics aim to approximate sound qualities of loudness, sharpness, and tonal loudness using acoustic data with a much lower time resolution. As such, the metrics are expected to be primarily applicable to sounds featuring relatively slow modulations in intensity. A particular application case is used to validate the metrics: sound quality predictions for unmanned aircraft systems.

The Sottek Hearing Model provides a means to analyse perceptual sound qualities including loudness, tonality, roughness, and fluctuation strength. An extension to the loudness algorithms is described, which addresses cases in which a source of interest is partially masked by other sounds, such as an encompassing ambient acoustic environment. The proposed extension is validated with objective predictions for unmanned aircraft systems (UAS) and artificial noise, and subjective test results for UAS. The partial loudness extension proposed also enables the calculation of further partial sound qualities, including partial tonality, partial tonal loudness, and partial sharpness. Software implementations of the tools are provided.

A method for the synthesis and auralisation of multirotor unmanned aerial systems (UAS) overflights is presented. The proposed method allows tracking of the spectral features of the recorded sound generated by an aircraft under a static flight operation. Then, synthesised signals for constant speed overflight are generated by applying spectral modifications accounting for the shifts in rotor rotational speeds. The auralisation process includes sound propagation modelling accounting for the conditions of an outdoor environment. Comparisons between recordings and auralised flight operations show that the proposed method can successfully reproduce the audible stimuli of small multirotor aircraft. The accuracy of the auralisations has been assessed based on sound pressure levels and sound quality metrics. The method presented achieved excellent agreement between the auralised signals and recordings for four aircraft varying in size, takeoff mass, and number of rotors. The applied method contributes to the generation of auditory stimuli that capture UAS spectral features in static flyovers and simulate forward flight operations in realistic outdoor flight scenarios. Building on this capability, the presented auralisation method adds to the set of tools available for assessing human responses to UAS noise.

This study explores the soundscape impact of unmanned aircraft systems (UAS) noise in an urban park using soundwalks in two urban park areas: one with prominent traffic noise from a nearby thoroughfare, and the other behind a row of tall pine trees with much diminished traffic noise and more prominent sounds of nature. Objective measurements and subjective responses reveal that takeoff and landing events have greater impact than flyovers, with stronger effects at lower altitudes. UAS noise significantly affects pleasantness and eventfulness in the quieter area, while the noisier area shows smaller perceptual changes. Despite greater differences in the quieter area, pleasantness levels remained higher than in the noisier, suggesting a compensatory effect of ambient noise. Detectability and tonality negatively correlated with pleasantness changes, while impulsiveness correlated positively with absolute pleasantness, likely due to ambient scene characteristics rather than UAS noise itself. The study highlights the importance of considering the background ambience in moderating the perception of UAS noise.

The classification of acoustic scenes and events is an emerging area of research in the field of machine listening. Most of the research conducted so far uses spectral features extracted from monaural or stereophonic audio rather than spatial features extracted from multichannel recordings. This is partly due to the lack thus far of a substantial body of spatial recordings of acoustic scenes. This paper formally introduces EigenScape, a new database of fourth-order Ambisonic recordings of eight different acoustic scene classes. The potential applications of a spatial machine listening system are discussed before detailed information on the recording process and dataset are provided. A baseline spatial classification system using directional audio coding (DirAC) techniques is detailed and results from this classifier are presented. The classifier is shown to give good overall scene classification accuracy across the dataset, with 7 of 8 scenes being classified with an accuracy of greater than 60% with an 11% improvement in overall accuracy compared to use of Mel-frequency cepstral coefficient (MFCC) features. Further analysis of the results shows potential improvements to the classifier. It is concluded that the results validate the new database and show that spatial features can characterise acoustic scenes and as such are worthy of further investigation.