Kolloquium - Details zum Vortrag
Sie werden über Vorträge rechtzeitig per E-Mail informiert werden, wenn Sie den Newsletter des kommunikationstechnischen Kolloquiums abonnieren.
Master-Vortrag: Continuous Acquisition of Individual HRTFs in Various Environments
Montag, 8. Januar 2018
Hörsaal 4G IKS
With the rise of binaural audio content, the demand for individual Head-Related Transfer Functions (HRTFs) becomes more pressing. Numerous measurement techniques have been proposed. Most of them require anechoic environments and assume that measurement subjects are perfectly still during rotation. In this thesis, it is investigated how conditions change if the subject can rotate self-determinedly during measurement and if the environment is reverberant. Emphasis is given to the influence of excitation signals using theoretical analyses and computer simulations. Several acoustic measurements have been conducted in various environments. Moreover, different head-tracking systems were included in the measurement setup in order to capture the exact relative position of sound source and subject.
Regarding excitation signals, this work points out complex trade-offs impacting Signal-to-Noise Ratio (SNR), distortion artifacts and spatial resolution of the acquired HRTF sets. To further optimize the choice of parameters, more research in psychoacoustics is necessary to specify requirements for naturally sounding HRTFs objectively. As a result of this thesis, the equivalence of two system identification algorithms, the Normalized Least-Mean-Square (NLMS) algorithm for arbitrary periodic excitation and Inverse Cyclic Convolution (ICC), has been proven for the first time. Furthermore, a new excitation signal has been introduced and a new head-tracking concept is presented. It has turned out to be very suitable for the application in HRTF measurements and provides promissing performance. As a consequence of freely moving subjects, additional post-processing is required to compensate for translational motion which unavoidably occurs with humans.
Alle Interessierten sind herzlich eingeladen, eine Anmeldung ist nicht erforderlich.