Skip to main content
Voice typing gets most words right on its own. Names, product terms and company jargon are harder, so Granola has three ways to help. You’ll find all of them in Settings > Voice typing, under Personalization.

​
Dictionary

Add names and terms you want spelled a particular way, like “Siobhan”, “Kubernetes” or “Projekt DolFin”.
  1. Click your workspace name in the bottom-left corner, then select Settings.
  2. Open Voice typing.
  3. Under Dictionary, type a term and press Enter.
Each term can be up to 50 characters and six words. This dictionary is only for dictation. It’s separate from the jargon list for meeting transcription, so add a term to both if you need it in each. The dictionary helps Granola recognize a word when you say it. It can’t replace one word with another, so adding “Granola” won’t turn “banana” into “Granola”.

​
Learn from your corrections

When you fix a misspelled name or term straight after dictating, Granola can add the corrected spelling to your dictionary for next time. The pill at the bottom of your screen tells you when a word is added, and lets you undo it. Only spelling fixes for names and terms are learned. Rewording a sentence doesn’t change your dictionary. Learned words appear in your Dictionary list, and you can remove them like any other term. To turn this off, go to Settings > Voice typing and switch Learn from your corrections off.

​
Screen context

With Screen context on, Granola reads text from the app you’re dictating into to spell names correctly, like the recipient’s name in an email or people in a Slack thread. It also lets Granola match the app:
  • In email, it separates the greeting and sign-off and groups your thoughts into paragraphs.
  • In chat apps like Slack, it keeps messages conversational and short.
  • In code editors and terminals, it keeps technical terms and their exact case.
  • In one-line fields, like a subject line or search box, it doesn’t add line breaks or a final period.
  • When you’re continuing a sentence, it fits in with the text around your cursor.
Screen context is on by default. Granola reads text, not screenshots, and it skips password fields. See Dictation history and privacy for exactly what’s read and where it goes. Dictation styles rely on screen context to know which app you’re in. If you turn screen context off, styles are turned off too.