After the training on speaker recognition, subjects were given a task in which speech was presented in noise; the task was to report what the speech was. There was an interesting correlation between the talker recognition results and the speech identification results.