Abstracts: Lecture presentation

Stephen Siegel (Fulcrum Acoustics, LLC)

A Modern Approach to Coaxial Loudspeaker Drivers for Sound Reinforcement Applications

Since the advent of the Altec 601 duplex (coaxial) loudspeaker in 1943, coaxial loudspeaker drivers have had niche appeal. Initial popularity was confined to studio monitors and home hi-fi speakers. Within the past decade, there has been a strong trend toward coaxial-based stage monitoring systems, but until recently no company had created a flexible range of coaxial solutions applicable to general sound reinforcement applications. We will discuss how coaxial drivers can uniquely important to sound reinforcement systems. We will demonstrate a coaxial-based system incorporating the discussed technologies.

Shoken Kaneko (Reseaerch & Development Divison, Yamaha Corporation)

A new binaural processing for realistic 3D audio applications

We present a method for modeling human ear shapes, and particularly, a method for obtaining a generic non-individualized head-related transfer function (HRTF), based on the arithmetic mean of human ear shapes.

The shape-based average HRTF is calculated from this average human ear shape with the boundary element method (BEM). The obtained average HRTF was evaluated by subjective experiments, revealing improved localization precision over a HRTF calculated from the shape of a mannequin head. Our approach does not require any measurements of HRTFs of the listener or selections of fitting HRTFs from a predefined database, and thus it can be practically utilized in any 3D audio or acoustic VR/AR application which makes use of HRTFs.

Inyong Choi (Department of Communication Sciences and Disorders, University of Iowa)

Cochlear implant users' real-world communication

Cochlear implants (CI) are the most successful neural prosthesis having transformed hearing function for more than 300,000 people worldwide. However, there is large variability in CI performance in real life, especially when users try to recognize speech and other sounds in noisy environments, with no clear understanding of which cortical mechanisms predict success. This presentation explores objective methods to assess the ability of CI users to detect complex auditory objects including speech in simulations of real-life environments, and to demonstrate the brain substrates for this. This presentation also introduces a newly developed neurofeedback training system that can strengthen CI users' cortical neural networks controlling speech-in-noise understanding.

Marc Ropez & Alan Macpherson (Yamaha Corporation of America)

History of Yamaha Pro Audio

Abstract will be updated soon

Zhiyao Duan (University of Rochester)

Retrieving Sounds Through Vocal Imitation

Searching sounds with text labels is often problematic as text labels do not often describe the detailed audio content. In this project we propose a novel approach for sound retrieval through vocal imitation. Vocal imitation is commonly used in human communication and can be employed for novel human-computer interaction. We propose both a supervised and an unsupervised system for this task. The supervised system trains a multi-class classifier using training vocal imitations of different sounds in a library and recognizes a new imitation query as being associated to one of the sounds. This system can thus only retrieve sounds that have training vocal imitations. The unsupervised system is more flexible in that it measures the feature distance between the imitation query and each sound in the library and returns sounds that are most similar to the query. The main challenge of designing these systems is finding an appropriate feature representation for the vocal imitations and sounds. Existing hand-crafted audio features do not work well given the variety of the sounds and vocal imitations as well as their surface-level mismatch. We propose to learn feature representations from training vocal imitations automatically using deep neural networks. Experiments show that these automatically learned features significantly outperform those carefully handcrafted ones in both supervised and unsupervised settings in sound retrieval.

Song Hui Chon (Rochester Institute of Technology)

Effect of Room Acoustics and Music Type on Listeners’ Preference of String Quartet Performances Recorded in Virtual Acoustics

Our earlier study reported that musicians enjoyed performing in virtually enhanced acoustic conditions significantly more than in unaltered room acoustics. The goal of the present study is to investigate whether recordings made in these more enjoyable performing conditions were also more enjoyable to the listeners, who were unaware of the acoustic conditions of the recordings. A listening experiment was carried out with nineteen musicians using a 2AFC paradigm. Stimuli were short segments recorded by five string quartet groups in three acoustic conditions, including two with virtual enhancements. Listeners indicated which segment of a given pair they preferred. Although ANOVA found no significant effect of condition on listeners’ preference, Bradley-Terry-Luce analysis reveals an interaction of the type of music and the acoustic condition.

Dave Anderson (University of Rochester)

Recent Developments in Driver Array-Based Flat Panel Loudspeaker Technology

Flat Panel Loudspeakers radiate sound by means of bending waves, and tend to have poor acoustic performance when compared with traditional loudspeakers. A majority of the problems can be mitigated by using driver arrays with additional electronic signal processing. Modal Crossover Networks are one use of this technique, employing arrays of driving elements to tune out undesirable vibrational characteristics. This method has been shown to drastically improve the low-frequency acoustic response and spatial coverage. The mid- and high-frequency regions of operation can still have irregular frequency response and spatial coverage characteristics using this method, and experimental evidence indicates that the characteristics of the panel materials dominate the performance in these regions. Important limiting factors in the design of such loudspeakers will also be discussed.