Tips and best practices

Sound quality

Speech recognition is more likely to be accurate if you work in a quiet place. Try to keep the background noise to a minimum when you're working.

A high-quality microphone is essential; follow the manufacturer's advice on using the microphone.

Test the sound produced by your microphone and device combination by recording your voice in a recorder app and listening to it. Is the sound clear? Do you think it's good enough?

Speaking

There is no need to over enunciate; speak continuously in a normal and clear way (not in short bursts).

If you have a cold, try to speak as normally as possible.

If you experience problems with recognition, don't speak more loudly or more slowly; continue speaking at your normal cadence.

Try to avoid breathing directly into your microphone, voicing hesitations (um/ah), long pauses in the middle of phrases and superfluous speech (thank you this is the end of the report).

Make sure to say punctuation marks.

Understanding speech recognition

Speech recognition is the computerized recognition of spoken words. Speech is transformed into text, this means written out, on the screen. This text is the recognition result.

Recognition of voice commands means that the program responds to the action rather than writing out the words spoken.

The system automatically formats some types of recognition results. Find out what to say to get automatic formatting.

The Manage Vocabulary page enables you to improve speech recognition by explicitly tailoring the system to your requirements. This might be necessary if words or phrases are repeatedly misrecognized. You can add or delete words and phrases. Any changes you make are only available to you; other users of the system aren't affected.