Speech Parameter Extraction by a Cooperative Feature and Pattern Selector Approach