How to Build a Speech Recognition Application: Second Edition: A Style Guide for Telephony Dialogues

We have also rewritten and expanded on Chapter 11, "Usability Testing and Performance Reporting." Most of the new material summarizes our experiences since the first edition was published. In addition, we have added a major section to Chapter 8 on "Challenges in Implementing Natural Language Systems"-new material that has been compiled since the first edition. We have also updated the yes-no discussion in Section 6.3 with a description of some recently-proven methods for handling mixed-initiative replies. Other improvements in this edition include updated dialogue examples, more copious cross references, and improvements to the index and glossary.

Since this book was published, many speech vendors and application developers have conducted formal studies and/or focus groups to learn more about design decisions and to provide guidelines for future voice user interface development. In all of the studies we are aware of, our empirical observations have been validated. Hence none of the guidelines in the first edition have been overturned or found to be in error. In several cases, we have upgraded our guidelines from "Good Practice" to "Recommended" or to "Required" based on these results.

Finally, we tried not to make any structural changes to this text. None of the section numbers have been changed except in Chapter 11, Appendix A, and the beginning of Chapter 9. So if people on your design team have copies of both the first and second editions, you should be able to cross reference them easily.