Dictation software allows you to create documents (email, wordprocessing, forms) by speaking instead of typing. Most dictation software allows you to speak in a natural way (conversational). To work effectively, these systems need to utilize databases of words and sounds (a lexicon). An initial training session is used to help the computer learn to interpret your voice and speech characteristics. This training (called "enrollement") will improve the accuracy of the dictation.

Important Things to Consider in Dictation

When dictating, try not to look at the computer screen since this may distract you from your thoughts and cause you to loose concentration. The computer will attempt to transcribe everything you say. So, make sure you think about it first before saying it.

The process for converting your thoughts into text is very different from typing your thoughts than when you're talking to a computer. In order to maximize the accuracy of the speech recognition software, the speech program will attempt to adjust the spellings of words based on the context of the sentence that it hears. In other words, speech programs analyze entire phrases or sentences, not just individual words. This is why you will often see nothing on the monitor/screen until just after you stop speaking. It takes a few seconds for the computer to figure out what you were saying.

If you don't think about what you are going to say before you say it, you may end up just blabbering on about the topic but not really saying anything important. If you were to stop in mid sentence, this would reduce the ability of the software to accruately transcribe what it heard.

Therefore, dictation into a computer will work most efficiently when you learn to form each sentence in your head before speaking it.

Pronounciation (052.htm)
Speak in a clear voice and pronounce each word. Speech recognition software is not perfect and may make mistakes in the translation of your voice.

Since the job of the speech engine is to convert your speech into text, the speech engine is a very good speller and will create text that does not have spelling errors. However, the words chosen by the speech engine may not be exactly what you had intended. You need to carefully read the text that the computer generates. For example, you may say: "You can recognize speech." and the computer would transcribe this into: "You can wreck a nice beach." These types of mistakes may be made by the computer. So check the text generated by the computer carefully.

Achieving Good Recognition Accuracy

1. Use a noise cancelling microphone
2. Use a digital microphone instead of an analog one
3. Make sure background noise is at a minimum
4. Perform the voice training / user enrollment
5. Practice at first by reading from a printed source (e.g., newspaper)
6. Speak clearly and slowly
7. Think about what you want to say before you say it
8. Train individual words that are not recognized

Commercial Solutions

Health Care Solutions