Speech recognition (also known as voice recognition) is the process of converting spoken words into computer text. The user speaks into a microphone and the computer creates a text file of the words they have spoken.
Although the accuracy of these systems has improved in the 21st century, they are still far from perfect. If you only need them to recognise a few words, for example the words one, two, three etc., then speech recognition software can be very accurate. This is why this feature has started to appear on modern mobile phones. But systems that can recognise any word spoken by any person in any language are still a few years away.
You can greatly improve the accuracy of these systems by training them to recognise just one voice. This is usually done by getting the user to read a short set piece of text into the microphone. The user looks at the text which has been created and corrects any mistakes the software has made. By repeating this a number of times the computer's accuracy increases a great deal.