Listening for people: Exploiting the spectral structure of speech to robustly perceive the presence of people

Abstract

As the desire to see robots ubiquitous in society grows, so does the need for providing the robots with the means of building awareness of any humans with which it may be sharing the environment. This paper presents a real-world suitable system which enables robots to robustly perceive the presence of people acoustically. The proposed binaural system first identifies voiced signal by means of a novel approach to Voice Activity Detection that exploits the spectral signature and characteristics of speech without reliance on a priori knowledge. Bearing estimates for each speaker are then made using a multitrack particle filter with a belief update function comprised of a Cross-correlation bearing estimate and an estimate of the speaker's fundamental frequency. Results, from an evaluation of each of the major system components and a system evaluation in which the robot successfully built human-centric situational awareness of the three humans with which it shared an office lunch-room containing typical background noises, are presented and discussed.

A survey on sound source localization in robotics: From binaural to array processing methods

“…One can cite for instance [80], or [81] where the PhaT processor is exploited on a 24 evenly spaced microphones array fitted on the 3.2 m-long walls of a room which is visited by a tour-guide robot. More recent use of the PhaT approach can be cited: [82] where a triangular 3-microphone array is used to infer source location from short time observations so as to cope with the movement of the sound source or the robot; [83] presents an evaluation of various real-time sound localization approaches from a cubical 8-microphone array in which GCC-PhaT is compared with beamforming techniques; [84] proposes a robust approach to the acoustic perception of the presence of people from a pair of microphones. But GCC-PhaT only takes into account the phase of the perceived signals in the intercorrelation computation, giving the same importance to each frequency.…”

“…This basic geometric rule is used in [90], the computed azimuths being involved into a neural network based sound source tracker. The same strategy is used in [78], or [84]. Following the same lines, one can deduce the cartesian coordinates r s = (u, v, w) of a source from the known positions r m n = (x n , y n , z n ), n = 1, .…”

Development of a Test-Bench for Evaluating the Embedded Implementation of the Improved Elephant Herding Optimization Algorithm Applied to Energy-Based Acoustic Localization

“…Additionally, in [55], an evaluation of real-time sound localization approaches are compared using an 8-microphone array. Similar architectures were presented in [56,57], where the common need of an array of microphones can be considered as a drawback for practical applications [58,59]. As such, the present work considers the acoustic energy-based model of [44].…”

A Robot Centric Perspective on the HRI Paradigm

“…This approach to person detection (presented in Hilsenbeck & Kirchner, 2011) is not 3D data reliant, but rather, is audio based. In this binaural approach the robot exploits the spectral structure (in the audio frequency domain) and characteristics specific to human-originating sound (speech or humming for instance) to robustly detect the presence and gender of human speakers.…”