Assessment of modal density and free path distribution in central-planned halls.

Central-planned halls are highly widespread in the historical architectures of the Western world, such as rotundae, Christian baptisteries, and Roman tombs. In such halls, whispering galleries, flutter echoes, and sound focusing are the acoustic phenomena mainly investigated by scholars. Instead, modal behaviour and free path distribution are generally less treated in literature. The present study explores the modal density at low frequencies and the relationship with the most recurrent free path lengths in three historical nearly circular spaces, here assessed as case studies. Acoustic measurements allowed the collection of objective experimental data, i.e., room impulse responses and the resulting room acoustics criteria. Wave-based numerical models allowed for the investigation of the eigenfrequencies distribution, while the free paths trend has been experienced through ray-based models. The main outcomes of both analyses show the prominence of the circular modes, rather than the diametral and the elevation ones. Moreover, the mean free path calculated using ray-tracing proves to be higher than the theoretical value commonly assumed for any kind of shape. The consequent longer reverberations compared to halls with other shapes and the same volume justify the significant support historically provided to sound signals by circular halls.

Singing in different performance spaces: The effect of room acoustics on singers' perception.

Classical singers' performances vary across different acoustic environments. The changes in the delivery are influenced by the singer's perception of the venue's acoustics. This study investigated these relationships using nine professional or semi-professional classical singers. Participants performed Giordani's "Caro mio ben" aria in five venues, and the acoustic parameters reverberance (T30 and EDT), clarity (C80), early vocal support (STv), and tonal color (EDTf) were measured. From a factor analysis of the subjective analysis three major factors emerged that, we propose, would represent three generalized percepts of Room Supportiveness, Room Noiselessness, and Room Timbre. These percepts correlated significantly with objective acoustic parameters traditionally linked to vocal support, reverberation, and timbre. Room Supportiveness and Room Noiselessness significantly contributed to the singers' likability of the acoustic environment, while Room Timbre did not. This indicates that singers' perceptual preference for a performance space may be influenced by factors affecting both auditory feedback and vocal function. These findings underscore the need for performing space designers to consider the unique needs of all stakeholders, including listeners and performers. The study contributes to the bridging of the gap between subjective perceptions and objective measurements, providing valuable insights for acoustic design considerations.

Blind source separation by long-term monitoring: A variational autoencoder to validate the clustering analysis.

Noise exposure influences the comfort and well-being of people in several contexts, such as work or learning environments. For instance, in offices, different kind of noises can increase or drop the employees' productivity. Thus, the ability of separating sound sources in real contexts plays a key role in assessing sound environments. Long-term monitoring provide large amounts of data that can be analyzed through machine and deep learning algorithms. Based on previous works, an entire working day was recorded through a sound level meter. Both sound pressure levels and the digital audio recording were collected. Then, a dual clustering analysis was carried out to separate the two main sound sources experienced by workers: traffic and speech noises. The first method exploited the occurrences of sound pressure levels via Gaussian mixture model and K-means clustering. The second analysis performed a semi-supervised deep clustering analyzing the latent space of a variational autoencoder. Results show that both approaches were able to separate the sound sources. Spectral matching and the latent space of the variational autoencoder validated the assumptions underlying the proposed clustering methods.

Feasibility of a finite-difference time-domain model in large-scale acoustic simulations.

Wave-based techniques for room acoustics simulations are commonly applied to low frequency analysis and small-sized simplified environments. The constraints are generally the inherent computational cost and the challenging implementation of proper complex boundary conditions. Nevertheless, the application field of wave-based simulation methods has been extended in the latest research decades. With the aim of testing this potential, this work investigates the feasibility of a finite-difference time-domain (FDTD) code simulating large non-trivial geometries in wide frequency ranges. A representative sample of large coupled-volume opera houses allowed demonstration of the capability of the selected FDTD model to tackle such composite geometries up to 4 kHz. For such a demanding task, efficient calculation schemes and frequency-dependent boundary admittances are implemented in the simulation framework. The results of in situ acoustic measurements were used as benchmarks during the calibration process of three-dimensional virtual models. In parallel, acoustic simulations performed on the same halls through standard ray-tracing techniques enabled a systematic comparison between the two numerical approaches highlighting significant differences in terms of input data. The ability of the FDTD code to detect the typical acoustic scenarios occurring in coupled-volume halls is confirmed through multi-slope decay analysis and impulse responses' spectral content.

Unsupervised analysis of background noise sources in active offices.

Inside open-plan offices, background noise affects the workers' comfort, influencing their productivity. Recent approaches identify three main source categories: mechanical sources (air conditioning equipment, office devices, etc.), outdoor traffic noise, and human sources (speech). Whereas the first two groups are taken into account by technical specifications, human noise is still often neglected. The present paper proposes two procedures, based on machine-learning techniques, to identify the human and mechanical noise sources during working hours. Two unsupervised clustering methods, specifically the Gaussian mixture model and K-means clustering, were used to separate the recorded sound pressure levels that were recorded while finding the candidate models. Thus, the clustering validation was used to find the number of sound sources within the office and, then, statistical and metrical features were used to label the sources. The results were compared with the common parameters used in noise monitoring in offices, i.e., the equivalent continuous and 90th percentile levels. The spectra obtained by the two algorithms match with the expected shapes of human speech and mechanical noise tendencies. The outcomes validate the robustness and reliability of these procedures.

Anechoic recordings of Italian opera played by orchestra, choir, and soloists.

Auralisation-based research needs anechoic recordings. Where opera is concerned, anechoic material is limited to a few recordings. Using the approaches of previous recordings, the present work shows methods and results of a recording set focused on the Italian melodrama period. The chosen motifs come from scores of Donizetti, Verdi, and Puccini in order to consider the most performed composers and the peculiarities of their styles, which span from the early 19th century to the 20th century. Excerpts from three operas have been played by professional musicians and soloists with an array of twelve microphones in a damped room, which can be considered anechoic. The multi-track files of the recorded materials are freely available for academic purposes.

Extraction of the envelope from impulse responses using pre-processed energy detection for early decay estimation.

The Schroeder's backward integration method and its applications have been widely studied in the literature; some papers analyze the performance of the method, some others suggest various enhancement techniques. In spite of these findings, there exist several cases where the energy decay curve extracted using the classical backward integration method and the parameters computed from it seem not always representative of the phenomenon under study. Among them, the cases where the early decay is dependent on strong, distinct reflections occurring just after the direct wave, as in most Italian opera houses. Other cases are measured impulse responses with a very low signal-to-noise ratio or missing the direct wave. In the literature, alternatives to the Schroeder's method have been proposed, ranging from Hilbert transform to non-linear processing techniques. In this work a method for the extraction of the envelope based on pre-processed energy detection for early decay estimation is proposed. It is shown that it returns an envelope well matching the first part of the decay even in non-linear cases, returning detailed information on the first part of the decay. The performance of the proposed method is presented and discussed for some exemplary impulse responses measured in historical opera houses. A preliminary study on the perceptive relevance of the method is finally presented.

A comparison of methods to compute the "effective duration" of the autocorrelation function and an alternative proposal.

The "effective duration" of the autocorrelation function (ACF), τ(e), is an important factor in architectural and musical acoustics. For a general application, an accurate evaluation of τ(e) is relevant. This paper is focused to the methods for the extraction of τ(e) values from the ACF. Various methods have been proposed in literature for the extraction of the τ(e) from a given signal, but these methods are not unambiguously defined or may not work properly in case of particular signals. Therefore, the general use of these methods may sometimes give rise to questionable results. In the present work, the methods existing in literature for extracting τ(e) are analyzed, their advantages and drawbacks are summarized, and finally an alternative method is proposed. The proposed algorithm is compared to those found in previous literature, applying them on the same sound signals (classic literature references and other ones publicly available on the Internet). It is shown that the results obtained with the proposed method are consistent with the results of the previous literature; moreover the proposed method may overcome some of the limitations of the existing methods.