Designing Effective Speech Interfaces

Continuous Speech Recognition

Continuous speech recognition systems allow the user to talk to the system without stops and pauses. Continuous speech recognition systems can recognize more utterances than a command-and-control system. The guidelines for continuous speech recognition differ somewhat from those for command-and-control.

Use “I”

In a continuous speech system, humans can speak in a continuous stream. Susan Boyce's research (1999) shows that when the system speaks using the first person, “I,” users report higher satisfaction with the system. The use of “I” in a continuous recognition system does not degrade user performance.

When the system speaks using the first person, “I,” users report higher satisfaction with the system.

Simulate a Conversation

When two people converse, they take turns talking and listening. A speech interface should do the same. When the computer finishes speaking, it should pause and wait for the human to respond.

Avoid Modal Interference

If the user is speaking to another person, for instance, a customer, and is then required to also speak to the computer, this produces modal interference. Modal interference will cause the user to be distracted and confused.