This deliverable covers the entire set of acoustic signal processing algorithms as developed for the given robot audition scenario and comprises software in addition to the report. More specifically, it describes the findings for
• Acoustic source and environment mapping and tracking for real-world scenarios. Following a state-or-the-art review with analysis and the baseline algorithms, the advance beyond the state of the art in algorithms for acoustic source and environment mapping will be presented, together with tracking methods for the robot scenario relevant in EARS and with a special emphasis on the novel and time-varying array geometries. Spatial filtering (T2.4). Spatial filtering methods for
microphone arrays in robots. Evaluation methodology description as well as results will be provided for prototype microphone arrays of the Nao head.
• Acoustic echo cancellation (T2.5). This reports acoustic echo cancellation for the special case in which we consider the moving robot with integrated and/or prototype microphone arrays.
• Multichannel noise reduction (T2.6). This will present the outcomes of the research on noise and interference suppression exploiting multichannel processing from the prototype microphone arrays for the Nao robot.
• Dereverberation (T2.7). The research on dereverberation will be reported in this deliverable and the relevant software implementations will be cross-referenced. It will include a state-of-the-art review, the methodology of evaluation and the results for baseline approaches set against the progress made beyond the state of the art
in the new developed algorithms. For all areas relevant software will be cross-referenced for the developed algorithms and evaluation results reported.
This Deliverable is due in M30 and has been submitted to the EC on June 30, 2016.
The corresponding software for the final prototype is provided by a repository (for EARS members) and the current version (June 30, 2016) can also be download as zip-archive.