Objective

Speech is produced using the lungs, vocal tract, and mouth making it so that each and every one of us have a unique voice. In this activity you will get experience with speech recognition software that turns speech into text. To learn more, check out some background information on Speech Recognition.

Equipment

Computer with WordDragon Naturally Speaking Preferred 10
Headset with Microphones
Calculator
A Paragraph (something to read out loud)

Try it

Option 1: Use Dragon without training. Speech-to-text will not be as accurate, but may be more robust to different speakers.

Option 2: Use short training on one student from the group. If the you do not have good diction, it will be frustrating and take more than 5 minutes. The software will have to be uninstalled (10 mins) and reinstalled (20 mins) before a different person can use it. Performance will be better for those who did the training, and worse for the others .

Steps

Step 1. Each person takes a turn using the headset and reading from a script. You will need to dictate punctuation (e.g., "comma," "period"). Girls should speak clearly and read at a normal pace, but it helps to pause between sentences or phrases to let Dragon catch up. It is OK if the you makes a mistake, or if the software makes a mistake. It will not help to yell at Dragon. Corrections could be done later.

Step 2. After each you have read, print the Word document. Count the number of correct words, and compute a percent accuracy. It may be funny to see what words were misunderstood.