Abstract: A loudspeaker array is described including several circular sound emitters that are bunked. One circular sound emitter can be either a circular array of loudspeakers (kind 1) or a toroidal loudspeaker (kind 2) The loudspeaker array can be composed of circular sound emitters of kind 1, kind 2 or both. A toroidal loudspeaker is a loudspeaker whose membrane is annular and whose enclosure has a central hole. Both toroidal loudspeaker and circular array radiate mainly in their normal axis, i.e., perpendicular to the plane comprising the membrane or the loudspeakers. Circular sound emitters are arranged like an end-fire array so that all the circular sound emitter centers are on the same axis. Each circular sound emitter is driven by its own signal, which results from a filtered version of the input signal, allowing the sound to be focused in the radiation direction.

Abstract: Method and system for real-time speech recognition is provided. The speech algorithm runs on a platform having an input-output processor and a plurality of processor units. The processor units operate substantially in parallel or sequentially to perform feature extraction and pattern matching. While the input-output processor creates a frame, the processor units execute the feature extraction and the pattern matching. Shared memory is provided for supporting the parallel operation.

Abstract: Method and system for real-time speech recognition is provided. The speech algorithm runs on a platform having an input-output processor and a plurality of processor units. The processor units operate substantially in parallel or sequentially to perform feature extraction and pattern matching. While the input-output processor creates a frame, the processor units execute the feature extraction and the pattern matching. Shared memory is provided for supporting the parallel operation.