TIDE supports usability engineering, evaluation and monitoring of spoken dialogue applications

Motivation & Project Description

• Implement an automatic testbed as an efficient tool for usability engineering, evaluation and monitoring of speech-based dialogue services.
• Usability is predicted in two steps:
  • Quantify the interaction between user and system in a parametric way.
  • Use parameters to predict usability, user satisfaction or acceptability – without directly asking the user!

Outcome:

• Testbed for predicting usability on speech-based dialogue platforms, including:
  • Algorithms for extracting parameters from the audio signals (signal level).
  • Algorithms for extracting parameters from the speech recognition hypothesis and from other logged information (symbolic level).
  • Usability prediction algorithms.