Table 4.

Questions To Ask a Speech System Vendor

Usability issues:

Is there a dictionary or language model available for my dictation context? Can the system learn new words from imported documents? Does the dictionary have specific abbreviations that I will use? Is the language model good enough to distinguish between commonly used tenses of a word? How easy is it to correct errors on the fly? Do I need to use the mouse or keyboard while correcting errors, or can I use voice alone? Can I create templates and macros to simplify complex dictation tasks? Can I use voice commands to navigate a document and correct errors? What is the baseline machine configuration I need for the system to keep up with my dictation speed—what speed processor, how much RAM, and how fast a hard disk? How large are the speech files, and how long will it take to load speech files when switching between users? How good is the dictation accuracy if my voice changes (e.g., if I have a cold)? Will I need to retrain for a short time if this happens? How clearly must I speak in order for the system to understand my voice (i.e., how much slurring does it tolerate)? How well is the system able to deal with pauses and stray sounds? (Must I speak continuously, or can I take breaks between sentences while I think?)

System issues:

Will I need to upgrade my system to use the speech recognition product? Can I use the product on my laptop system? What sound hardware and microphone do I need? Does the system permit dictation into other applications (e-mail systems, word processors)? Is there a mechanism for integrating the speech software with my medical record system? Can the system import data (medication lists, laboratory reports) from other applications? Can I dictate directly into a note field in my medical record system? Will voice commands work inside the medical record system? Can the system run in a network environment? Can I transport my speech files from one computer to another, or will I need to retrain on the new computer? How long will it take to download speech files from a central server (i.e., how large are the speech files)?