ABSTRACT
The simplest form of processing of an encoded signal involves a direct translation of the incoming ciphers into the final usable form. This suffices when the signal is relatively noise-free and unambiguous. If the signal cannot be taken at face value because it is noisy, which is to say, variable or ambiguous, then more complicated forms of processing are necessary. Any form of processing that relies heavily on the receiver's own internal knowledge sources and algorithms is called "top-down processing." Given the great amount of variability speech scientists have found between the physical parameters of speech and the ciphers which they assume are present in the signal, speech would seem to be a prime example of a signal requiring top-down processing. Nevertheless, in spite of the great amount of research done on this problem, it is still possible to raise questions about the degree of top-down processing necessary in speech.
