Correct all recognition errors. The system is constantly learning and if you do not correct errors, the way that the system learns will be counterproductive.
Constantly checking and correcting every small mistake as it occurs is not efficient working practice with speech recognition. We recommend waiting until you have finished, for example, a whole paragraph and then correct it.
You can correct by manually typing or by selecting the incorrect text and dictating over it.
For recognition errors we recommend manually typing the correct text.
For mistakes (for example, 'left arm' instead of 'right arm') we recommend selecting the text and dictating over it. This ensures that the sound remains synchronized with the text which means the system can learn from it.
When selecting text in order to dictate over it, it is better to select a whole phrase (the words surrounding the wrong word) and not individual words.
Dictate your information in the order you want it to be in the report, for example, findings come before conclusion. Try to avoid cutting and pasting sections of the text within the report.
