No abstract

No abstract

For sound localization methods to be useful in realtime scenarios, the processing power requirements must be low enough to allow real time processing of audio inputs. we propose a new binaural sound source localization technique based on using only two microphones placed inside the ear canal of a robot dummy head. The head is equipped with artificial ears and is mounted on a torso. In contrast to existing 3D sound source localization methods using microphone arrays, the novel method presented employs only two microphone and is based on a simple correlation approach using a generic set of HRTFs. The proposed method is demonstrated through simulation and is further tested in a household environment. This set up proves to be very noise-tolerant and is able to localize sound sources in free space with high precision.

An algorithm for real time humanoid sound localization and tracking using only two microphones in a highly reverberant environment is proposed. Several recently developed 3D humanoid sound localization algorithms require the environment to be anechoic. Also, the resolution of front-back ambiguity problem during sound localization requires the knowledge about the reference signals. Using HRTF based sound localization together with extended kalman filtering, we are able to accurately track moving sound sources in real time in a highly reverberant environment. This algorithm uses only two microphones and requires no prior knowledge of the reference signals.

No abstract

Telepresence is generally described as the feeling of being immersed in a remote environment, be it virtual or real. A multimodal telepresence environment, equipped with modalities such as vision, audition, and haptic, improves immersion and augments the overall perceptual presence. The present work focuses on acoustic telepresence at both the teleoperator and operator sites. On the teleoperator side, we build a novel binaural sound source localizer using generic Head Related Transfer Functions (HRTFs). This new localizer provides estimates for the direction of a single sound source given in terms of azimuth and elevation angles in free space by using only two microphones. It also uses an algorithm that is efficient compared to the currently known algorithms used in similar localization processes. On the operator side, the paper addresses the problem of spatially interpolating HRTFs for densely sampled high-fidelity 3D sound synthesis. In our telepresence application scenario the synthesized 3D sound is presented to the operator over headphones and shall achieve a high-fidelity acoustic immersion. Using measured HRTF data, we create interpolated HRTFs between the existing functions using a matrix-valued interpolation function. The comparison with existing interpolation methods reveals that our new method offers superior performance and is capable of achieving high-fidelity reconstructions of HRTFs. IntroductionTelepresence systems aim at supplying the senses of a human operator with stimuli that are perceptually plausible to an extent that the operator develops a persistent experience of actually being somewhere else, a so-called sense of presence. The most important stimuli are vision, audio, and haptics. The generic model of telepresence and teleaction is depicted in Figure 1. The perceptual world that the operator is experiencing is built up of sensory data that is provided by a teleoperator, that is, a robot, located at a remote site. At the local operator site, a human operator is interacting with a multimodal humanmachine interface that renders the sensory data. The human operator manipulates the teleoperator through the interface, which generates the corresponding control signals to be transmitted to the remote site. The integration of *Correspondence to keyrouz@tum.de.

We combine binaural sound-source localization and separation techniques for an effective deployment in humanoid-like robotic hearing systems. Relying on the concept of binaural hearing, where the human auditory 3D percepts are predominantly formed on the basis of the sound-pressure signals at the two eardrums, our robotic 3D localization system uses only two microphones placed inside the ear canals of a robot head equipped with artificial ears and mounted on a torso. The proposed localization algorithm exploits all the binaural cues encapsulated within the so-called Head Related Transfer Functions (HRTFs). Taking advantage of the sparse representations of the ear input signals, the 3D positions of two concurrent sound sources is extracted. The location of the sources is extracted after identifying which HRTFs they have been filtered with using a well-known self-splitting competitive learning clustering algorithm. Once the location of the sources are identified, they are separated using a generic HRTF dataset. Simulation results demonstrated highly accurate 3D localization of the two concurrent sound sources, and a very high Signal-to-Interference Ratio (SIR) for the separated sound signals.

For the purpose of a realistic rendering of moving sound with no noticeable discontinuity, binaural synthesis of the timevarying sound field is performed by updating a dense grid of head related transfer functions, HRTFs. Unless the differences in HRTFs are sufficiently small, a direct switching between them will cause an audible artifact that is heard as a click. To avoid this problem and ensure the availability of enough HRTFs, we have first reduced the HRTFs using PCA and then proposed a binaural impulse-response interpolation algorithm based on the solution for the rational minimal statespace interpolation problem. Compared with existing interpolation techniques, this method allowed very precise reconstruction of HRTFs in the horizontal plane and proved to have superior performance for a wide range of azimuths.