A method for decomposing speech into a small number of temporally overlapping events has been proposed by Atal. A survey is given of the stages in this method, illustrated by intermediate results at each step in analysing real speech. This reveals some problems at early stages in the original method developed by Atal. These problems are not fundamental to the philosophy of this apprqach, and some suggestions are made for overcoming them in future work .
|Journal||IPO Annual Progress Report|
|Publication status||Published - 1984|