Computer Science City College of New York
CSc 59866 - CAPSTONE I

Spoken Dialog Systems and Voice XML

Assignment 1 

  1. Test and analyze at least 2 spoken dialog applications:
  • Pain Diary:
    1. To register: Go to http://www.cunyspeechlab.com, choose ‘Pain Diary”, and then Enrollment.
    2. You will fill a form, and will receive an email with your password, and further instructions. You don’t have to enter your real personal data (age and ethnicity). You can choose there any random entry. Originally they were  required by NIH to guarantee that the study is balanced. 
    3. Place at least 1-2 call a day  (more is better!) over a period of a week – 10 days. Experiment with the system and document the dialogs.
    4. Write a report describing your perceived call flow (as a diagram); how the system behaves when it makes recognition errors; how it reacts to silence; does it offer help? How?; what happens if  you make a mistake and confirm negatively the right answer? Does it become easier/faster to complete a session as you get more experienced? Does it allow the users to change topic (mixed initiative)?
  • Tall free directory assistance.
    1. Call 1800-555-1212 after 5 pm.
    2. Place at least 10 calls. Try to find numbers for big companies such as airlines  or banks, and small companies – like pizza shops near you. Try to find the number for city college. Document the dialogs.
    3. Write a report describing your perceived call flow (as a diagram); how the system behaves when it makes recognition errors; how it reacts to silence; does it offer help? How?; what happens if  you make a mistake and confirm negatively the right answer? Does it become easier/faster to complete a session as you get more experienced? Does it allow the users to change topic (mixed initiative)?
  1. Research assignment (you can do it in groups). Explore two VoiceXML depelopment environments:

 

a)      http://cafe.bevocal.com/index.html

b)      http://www.vxml.org/ or IBM Voice Toolkit http://www-306.ibm.com/software/pervasive/voice_toolkit/

 

Create an account in both sites, and run the Hello world example.

Compare the two development environments. Which one do you prefer and why?