Applications of 3-Dimensional Spherical Transforms to Acoustics and Personalization of Head-related

Channel:
Microsoft Research
Subscribers:
344,000
Published on ● Video Link: https://www.youtube.com/watch?v=KqZejYbBYfQ


Duration: 55:59
1,668 views
22

The spherical harmonic transform (SHT), which returns spatial frequency components of data or distributions determined on the unit sphere, has found many applications in acoustics, such as spatial sound capture and reproduction, beamforming with spherical arrays, analysis of transducer radiation patterns, interpolation of head-related transfer functions (HRTFs) and others. However the SHT is a 2-dimensional transform, not suited for 3-dimensional data that vary not only with angle but additionally across the radial dimension. This work examines two 3-dimensional spherical transforms, namely the spherical Fourier-Bessel transform (SFBT) and the spherical harmonic oscillator transform (SHOT), and considers their potential uses in acoustics. The study presents preliminary results on their application to personalization of HRTFs, avoiding the cumbersome task of measuring them directly. Assuming that head-shape similarity correlates to some extent with HRTF similarity, we employ the aforementioned transforms to get a spectral representation of the user's head scan, and determine its distance from the spectra of head scans associated with the HRTF database.



Other Videos By Microsoft Research


2016-06-13Modern Deep Learning through Bayesian Eyes
2016-06-13Towards Understandable Neural Networks for High Level AI Tasks; Part 2
2016-06-13Towards Understandable Neural Networks for High Level AI Tasks - Part 4
2016-06-13What are the prospects for automatic theorem proving?
2016-06-13Towards Understandable Neural Networks for High Level AI Tasks - Part 3
2016-06-13Artist in Residence (formerly Studio99) Presents: Michael Gough and "Drawing as Literacy."
2016-06-13Towards Cross-fertilization Between Propositional Satisfiability and Data Mining
2016-06-13Making Objects Count: A Shape Analysis Framework for Proving Polynomial Time Termination
2016-06-13Human factors of software updates
2016-06-13Machine-Checked Correctness and Complexity of a Union-Find Implementation
2016-06-13Applications of 3-Dimensional Spherical Transforms to Acoustics and Personalization of Head-related
2016-06-13Network Protocols: Myths, Missteps, and Mysteries
2016-06-13Optimal and Adaptive Online Learning
2016-06-13Speaker Diarization: Optimal Clustering and Learning Speaker Embeddings
2016-06-13Multi-rate neural networks for efficient acoustic modeling
2016-06-13Unsupervised Latent Faults Detection in Data Centers
2016-06-13System and Toolchain Support for Reliable Intermittent Computing
2016-06-13Gates Foundation Presents: Crucial Areas of Fintech Innovation for the Bottom of the Pyramid
2016-06-13Social Computing Symposium 2016: Harassment, Threats, Trolling Online, Diversity in Gaming is Vital
2016-06-13Bringing Harmony Through AI and Economics
2016-06-13Approximating Integer Programming Problems by Partial Resampling


Tags:
microsoft research
spherical harmonic transform acoustics
hrtfs
computer systems and networking