Monday, May 22, 17:00 — 18:00
Juha Backman (Chair)
EB08-01 Motion-to-Sound Latency Measurement Procedure for VR Sound Reproduction
Jorgos Estrella (Presenting Author), Jan Plogsties (Author)
In the last couple of years, virtual auditory displays have finally reached the consumer market as part of emerging VR technologies. One of the challenges VR technology providers have to face is to reach affordable low motion-to-sound latency. Low latency is a very important factor while aiming towards immersive spatial sound reproduction. In this e-Brief a motion-to-sound latency measurement approach is proposed. This method employs a simplified parallel system running externally as reference. Here, a second head-orientation sensor is used to modulate a signal generator. Correlation analysis between the generated signal and the output signal of the device under test are used to assess latency.
Engineering Brief 345
EB08-02 Flexible Python Tool for Dynamic Binaural Synthesis Applications
Annika Neidhardt (Presenting Author), Thomas Köllmer (Author), Florian Klein (Author), Niklas Knoop (Author)
In this report, we present an open source tool for real-time dynamic binaural synthesis implemented in Python on top of PyAudio. The core is an efficient implementation of the uniformly partitioned convolution with the overlap-save approach. The dynamic and interactive reproduction of spatial audio scenes has become a common requirement in science and industry. Use cases are various, reaching from listening tests considering head rotation to complex reproduction scenarios for augmented or virtual reality, e.g., in combination with head mounted displays. With Python as a flexible and easy to learn programming language, PyBinSim offers great value in research and teaching of binaural synthesis. Source code, examples and documentation are available online.
Engineering Brief 346
EB08-03 A Self-Calibrating Earphone
Juha Backman (Presenting Author), Tom Campbell (Author), Marko Hiipakka (Author), Jari Kleimola (Author)
A self-calibrating system estimates the acoustical transfer function from sound pressures at the entrance of the ear canal to sound pressures at the eardrum: An earphone plays a broadband sound into the auditory meatus and an in-ear microphone then receives the sound at the entrance of the ear canal. Parenthetically, assessing calibration, results showed that spectral analysis of recordings of this signal is replicable to within 3 dB from 0.5 to 22 kHz for each given ear. A digital signal processing unit calculates an individualized filter from that signal. The calculated filter neutralizes the transfer function via software, which controls the digital signal processing unit’s output into the earphones whilst playing media.
Engineering Brief 347
EB08-04 End-To-End Process for HRTF Personalization
Tomi Huttunen (Presenting Author), Antti Vanne (Author)
The personalization of the head-related transfer functions (HRTFs) improves externalization and spatialization in headphone listening. The accurate measurement of an individual HRTF is time-consuming and complicated that has led to increased interest towards simulation based HRTF acquisition. The main challenge for simulations has been the lack of the fast and simple method to generate the three-dimensional (3D) geometry of the head and pinnae. On the other hand, a numerical solution of the 3D wave equation that characterizes the HRTF has been considered computationally demanding. We introduce an end-to-end process from the acquisition of the geometry to use of the personalized HRTFs in several applications. Results from the preliminary listening tests and future improvements are also discussed.
Engineering Brief 348