 |
Workshop on the Auditory Basis of Speech PerceptionKeele University, UK |
 |
|
Workshop on the Auditory Basis of Speech PerceptionKeele University, UK |
|
An architecture designed to combine separately operating estimators is described and evaluated. This architecture takes advantage of the constraints on the estimators to determine the accuracy of the estimates they produce, and combines the estimates based on their accuracies to produce a final estimate. Parameter values concerning the target being estimated and required by the estimators are determined from the final estimate and fed back to the preliminary estimators for use in the next processing frame. An implementation of the architecture is evaluated using a male target talker and female jammer talker under several spatial and target-to-jammer ratio (TJR) conditions. The implementation is able to yield improved TJR under unfavorable TJR, but does not do so consistently across TJR or spatial conditions. The architecture is discussed in terms of its relation to human auditory scene analysis and phenomena.
Bibliographic reference. Woods, William S. / Hansen, Martin / Wittkop, Thomas / Kollmeier, Birger (1996): "A scene analyzer for speech processing", In ABSP-1996, 232-235.