Open
Access
Intelligent Mixing Systems (IMS) are rapidly becoming integrated into music mixing and production workflows. The intelligences of a human mixer and IMS can be distinguished by their abilities to comprehend, assess, and appreciate context. Humans will factor context into decisions, particularly concerning the use and application of technologies. The utility of an IMS depends on both its affordances and the situation in which it is to be used. The appropriate use for conventional purposes, or its utility for misappropriation, is determined by the context. This study considers how context impacts mixing decisions and the use of technology, focusing on how the mixer’s understanding of context can inform the use of IMS, and how the use of IMS can aid in informing a mixer of different contexts.
Download: PDF (HIGH Res) (546KB)
Download: PDF (LOW Res) (165KB)
The development of intelligent music production tools has been of growing interest in recent years. Deep learning approaches have been shown as being a highly effective method for approximating individual audio effects. In this work, we propose an end-to-end deep neural network based on the Wave-U-Net to perform automatic mixing of drums. We follow an end-to-end approach where raw audio from the individual drum recordings is the input of the system and the waveform of the stereo mix is the output. We compare the system to existing machine learning approaches to intelligent drum mixing. Through a subjective listening test we explore the performance of these systems when processing various types of drum mixes. We report that the mixes generated by our model are virtually indistinguishable from professional human mixes while also outperforming previous intelligent mixing approaches.
Download: PDF (HIGH Res) (2.3MB)
Download: PDF (LOW Res) (508KB)
This two-part paper explores 3-D sound field reproduction by combining low order spherical ambisonics and high order circular ambisonics in order to provide higher resolution in the horizontal plane. Part I aims at rigorously establishing the solution of the Sound Field Reproduction (SFR) problem for a circular distribution of sources. This solution is derived from the cylindrical and spherical distributions to obtain two distinct approximate formulations involving either cylindrical harmonic or spherical harmonic expansions. Numerical simulations show that the two formulations lead to similar results. In the context of mixing a dense circular loudspeaker array for SFR in the horizontal plane with a sparse spherical array for 3-D SFR, the spherical harmonic solution for the circular source array will be privileged. It is then possible to combine the two systems in a high-pass/low-pass architecture in the spherical harmonics space. The combined system is described and investigated in Part II.
Download: PDF (HIGH Res) (10.5MB)
Download: PDF (LOW Res) (583KB)
This two-part paper explores 3-D sound field reproduction by combining low order spherical ambisonics and high order circular ambisonics in order to provide higher resolution in the horizontal plane. Part I is dedicated to rigorously establishing the solution of the Sound Field Reproduction (SFR) problem for a circular distribution of sources with either cylindrical harmonic or spherical harmonic expansions. In this second part the concept of spatial filtering is presented. A spatial low-pass filter using spherical harmonic expansion ensures proper reconstruction of elevation cues. Similarly a spatial high-pass filter using only higher orders of a circular harmonic expansion extends the zone of accurate reproduction in the horizontal plane. Numerical simulations of the hybrid system, based on the combination of the two spatial filters, are carried out and compared with the results of the classical spherical Higher-Order Ambisonic approach. The accurate reproduction area in the horizontal plane is larger in the case of the hybrid method than with the spherical approach, be it for sound pressure, direction-of-arrival, or sound field diffuseness. This method therefore makes it possible to extend the reconstruction zone in the horizontal plane by adding sources in the equator of a spherical array.
Download: PDF (HIGH Res) (19.9MB)
Download: PDF (LOW Res) (1.1MB)
An improved gradient descent method with multiple variable step sizes for identifying nonlinear parameters of moving-coil loudspeakers is proposed. This method monitors parameters trends during identification and adaptively multiplies or attenuates corresponding step sizes, eliminating the need for manual adjustments. Simulations and experiments reveal that it is as accurate as the fixed step size method and more efficient and robust against initial linear parameter errors.
Download: PDF (HIGH Res) (3.9MB)
Download: PDF (LOW Res) (506KB)
Binaural audio requires the use of crosstalk cancellation if reproduced using loudspeakers. A three-listener crosstalk cancellation system has been designed and built as part of this work. Simulations for different loudspeaker distributions are presented and a large system span is shown to increase low-frequency crosstalk cancellation performance while a denser loudspeaker distribution in front of a given listener increases mid to high-frequency crosstalk cancellation. The system’s performance under perturbation of the listeners and loudspeakers is investigated and at high frequencies loudspeakers further away from any given listener are shown to be affected more by perturbations than those nearer the listener. To this issue, a novel implementation of weighting loudspeaker source strengths using loudspeaker-dependent regularization is developed and optimized for use in this system. Hence, at high frequencies just loudspeakers close to the listener are used. This is shown to create a more robust solution than traditional crosstalk cancellation filter design when the system has undergone perturbations.
Download: PDF (HIGH Res) (6.3MB)
Download: PDF (LOW Res) (546KB)
Using digital signal processing techniques is an effective and low-cost way to compensate for nonlinear distortion of loudspeakers induced by the motor and suspension. Research shows that the nonlinear viscoelasticity of the suspension is also an important factor that distorts the acoustic output, but it is ignored in existing compensator designs. Therefore, in this paper, a new compensator that considers the nonlinear viscoelasticity of the suspension is proposed to reduce the nonlinear distortion. Experimental results demonstrate that the proposed compensator can significantly reduce the total harmonic distortion and intermodulation distortion in sound pressure while negligibly affecting the output sound pressure level.
Download: PDF (HIGH Res) (1.6MB)
Download: PDF (LOW Res) (727KB)
Measuring and controlling program loudness has become an important aspect of systems and operations in broadcast and online delivery. A panel of experts, chaired by David Bialik, gathered for a workshop at the recent convention to discuss the evolution of standards and the latest practices in this challenging field.
Download: PDF (301KB)
Live sound systems engineers discussed the challenges of optimizing installations remotely during a workshop presented at the Fall 2020 convention. A number of aspects of system alignment can be carried out in this way, provided there is suffi cient experience at both ends of the process. The more subjective elements of system setup really need to be done by the engineer who is using it on site.
Download: PDF (97KB)
