Thesis

Back to list

Automatic DOA estimator for spatial music recordings

Authors Maurer, S., Beck, S.
Year 2020
Thesis Type Audio Engineering project
Topic Audio Signal Processing
Keywords 3D sound, Ambisonics, B-format, Higher-Order Ambisonics (HOA), microphone arrays, signal processing, Spatial Audio, Software
Abstract The orientation in a room and its parameters are important knowledge when it comes to ambisonic musical performances. In this project an audio plug-in is developed which enables performers to estimate the position of their instruments in a room related to an ambisonic microphone array. Two different approaches of this task will be examined. Both are based on the same idea of building an intensity vector for determining the direction of arrival of the sound sources. One approach is to calculate the intensity vector of the whole array signal in the frequency domain which is filtered with a mask to separate the different instrument signals. The second approach is to calculate the intensity vector of an ambisonic room impulse response. The room impulse response is extracted from the musical signals with the help of a LMS-algorithm. With this room impulse response it is also possible to find room parameters like reverberation time and the dimensions of the room. Since it is a difficult task to make a stable and real time compatible algorithm for this, it will not find place in the plug-in but different methods to determine those parameters will be evaluated.
Supervisors Frank, M., Rudrich, D.