Binaural Reproduction of Signals captured in a reverberant Room with a Virtual Artificial Head
* Presenting author
As an alternative to conventional artificial heads, a microphone array-based filter-and-sum beamformer, referred to as Virtual Artificial Head (VAH), can be used to synthesize the directivity patterns of individual HRTFs (Rasumow et al. JAES 2017). Fallahi et al. (DAGA 2017/18) proposed to calculate the beamformer filter coefficients by solving a constrained optimization problem, i.e. by minimizing a narrowband least-squares cost function subject to constraints on the spectral magnitude synthesis error and the mean White Noise Gain. So far, this method has been evaluated only for simulated microphone arrays in free-field conditions. Results showed that, when choosing a proper array topology and appropriately defined constraints, the VAH perceptually outperforms classical artificial heads for music content in the horizontal plane (Fallahi et al. AES/Tokyo 2018) with regard to overall audio quality. In this contribution, real recordings were performed with a prototype VAH in a reverberant environment, where sensor noise and positional stability of the microphones come into play, and reflections from azimuths and elevations different from those of the source(s) are present. It is investigated to which extent the parameters and constraints within the optimization problem need to be modified in order to achieve a perceptually convincing binaural sound reproduction.