The Speech Systems Incorporated (SSI) commercial, large-vocabulary, speaker-independent, continuous speech recognition system is described. The system utilizes a novel approach to speech representation: a two-stage encoding of speech, with an intervening compression of acoustic frames (segmentation) between the encoding stages, and a linguistic decoding process suitable for large, variable-duration segments. Binary decision trees trained using the maximum mutual information (MMI) criterion serve as encoders. The features used in encoding are listed, and their ability to discriminate the phonetic content of the speech is analyzed. Recognition results are given for a speaker-independent continuous speech, grammar-constrained radiology reporting product, and for an isolated-word grammar of high perplexity.>

The notion of a qualified function is introduced as a general means of representing the parameters of dynamic systems. Two specific types of qualified functions are defined for the analysis of the behavior and performance of structured programs. Transformation functions represent the values of variables during execution and timing algorithms express the execution times of programs symbolically. Complete rules of derivation for transformation functions and timing algorithms are given for the control mechanisms of sequence, selection, fixed loop, and while statement. Deterministic and stochastic simplification of transformation functions and timing algorithms are investigated and methods of eliminating recursion for expressions corresponding to while statements are studied.