A computer’s voice recognition can be frustrating to deal with. Most applications, including Vista’s built-in voice recognition require training and customization.
Here’s a video of a Samsung Q1 owner using One Voice’s Media Center Communicator with no voice training. The demo shows the UMPC promptly playing songs and movies as the user commands, but I have no idea how large of a media library he’s working with. Would the commands work so smoothly if the application had to match his voice input against a library of thousands of songs?