An article by Chen and Chen (1995) compared the quality of two speech recognition systems. A benchmark of 2000 words was submitted to both the generalized minimal distortion segmentation (GMDS) system and the continuous density hidden Markov model (CDHMM). The following table contains the results of the performance of the two systems, the number of words correctly and incorrectly identified.

a. Are the proportions of correct identifications by the two systems correlated or independent? Justify your answer.
b. Test whether the proportions of correct identifications differ for the two systems. Use a 5 .05. Carefully state your conclusion in terms of the population parameters.
c. Construct a 95% confidence interval for the difference in the two systems proportions of correct identifications.

Transcribed Image Text:

CDHMM Correct Incorrect GMDS Correct Incorrect 1.921 16 58