Source ambiguity resolution of overlapped sounds in a multi-microphone room environment

Abstract

When several acoustic sources are simultaneously active in a meeting room scenario, and both the position of the sources and the identity of the time-overlapped sound classes have been estimated, the problem of assigning each source position to one of the sound classes still remains. This problem is found in the real-time system implemented in our smart-room, where it is assumed that up to two acoustic events may overlap in time and the source positions are relatively well separated in space. The position assignment system proposed in this work is based on fusion of model-based log-likelihood ratios obtained after carrying out several different partial source separations in parallel. To perform the separation, frequency-invariant null-steering beamformers, which can work with a small number of microphones, are used. The experimental results using all the six microphone arrays deployed in the room show a high assignment rate in our particular scenario.

Lightweight multi-DOA tracking of mobile speech sources

“…It, technically, carries out a basic form of sound separation throughout all possible directions and then 'decides' which directions are valid sound sources and which are not. A good example of this decision is presented in [37] where, assuming there are only two sound sources in a meeting room, the beamforming technique 'decides' which sound source is assigned to which sound source class. These decisions are carried out at specific intervals, with which several methodologies can be employed for tracking purposes, such as Kalman filters [18] or particle filtering [38].…”

Joint model-based recognition and localization of overlapped acoustic events using a set of distributed small microphone arrays

“…The steering beamformers at the front-end of the system are designed to work with the horizontal row of three microphones available for each of the arrays in the smart-room. An illustrative example of the beam pattern is shown in Figure 4 [22]. Note how the beams for the angle of interest are rather constant along frequency.…”

“…Figure 3 shows the smart-room layout with the position of its six T -shaped four-microphone arrays on the walls [22]. The steering beamformers at the front-end of the system are designed to work with the horizontal row of three microphones available for each of the For system training, development and testing, we have used the audio part of a multi-modal database recorded in that smart-room with all microphones [6] 1 .…”