In short
Spoken sentences often include false starts and repetition. Zack Typing’s text actions revise a transcript when you choose them: Improve smooths wording, Shorten condenses it and Professional adjusts the tone for work.
Original
Hi everyone, um, about the draft, me and Tom will send it this afternoon, so like, let us know by tomorrow if its okay.
Word for word, as you said it.
Improve
Hi everyone, Tom and I will send you the draft this afternoon. Let us know by tomorrow if it works for you.
No filler, reads like writing – in your tone.
Shorten
Tom and I will send the draft this afternoon – feedback by tomorrow, please.
Every point stays, in fewer words.
Professional
Hi everyone, Tom and I will send you the draft this afternoon. Please let us know by tomorrow whether it works for you.
Clear and polite, the request stays firm.
Choose by the result you need
| Action | Task |
|---|---|
| Improve | Revise awkward wording and transitions. |
| Shorten | Remove detours and condense the main points. |
| Professional | Make overly casual or abrupt wording more measured and polite. |
| Original | Return to the recognized transcript. |
Default instructions ask the model to retain the language and important details. Still check names, numbers and commitments, especially when shortening could remove something you meant to keep.
Choose while recording
Select a chip on the recording bar while you speak. After recognition, the action runs before the edited text is inserted. If it fails or takes too long, the app may use the original and marks that outcome.
You can choose another version later. An existing variant of the same dictation is reused, and Original remains available.
Custom instructions and costs
From Basic, edit names and instructions under Chips. Up to three text actions can be enabled at once, with Original available alongside them.
Each new action is another model request. With Zack Software, it needs internet access and credits. With your own provider, it uses your configured text model. The built-in offline speech model does not run text actions.
Related questions
Can I edit the text myself?
Yes. Open the dictation and change the relevant passage with the keyboard. For one incorrectly recognized name, this is often the quickest option.