Have you ever been frustrated by computer telephone systems that do not understand your voice commands? Quite a bit of effort goes into designing these systems to minimize voice-recognition errors. Researchers at the Oregon Graduate Institute of Science and Technology developed a new method of voice recognition (called a remapped network) that was thought to be an improvement over an existing neural network. The data shown are based on results of their research.

Does the evidence suggest that the remapped network has a different proportion of errors than the neural network? Use the α = 0.05 level of significance.

Remapped Network Recognized Word (success) Did Not Recognize Word (failure) Recognized Word (success) 9326 385 Neural Network Did Not 456 29 Recognize Word (failure)