Select-to-Transform

Transform any existing text using natural voice commands

Select-to-Transform lets you rewrite, reformat, or translate any text in any app — just select it, press the hotkey, speak an instruction, and the selection is replaced with the transformed result.

Select-to-Transform uses the LLM to transform the selected text according to your spoken instruction. It does not respond conversationally — it only returns the modified version of what you selected.

Workflow

1. Select the text you want to transform

   Highlight any text in any application — a sentence in an email, a paragraph in a document, a comment in your code editor. Any selection that can be copied via Ctrl+C will work.

2. Press Ctrl+Shift+K

   The recording overlay appears near your cursor, just like a normal dictation session. Trueears detects that you have text selected and switches into transform mode automatically.

3. Speak your transformation instruction

   Describe what you want done with the selected text. For example:

   • “Make it professional”
   • “Translate to Spanish”
   • “Convert to bullet points”
   • “Fix the grammar”
   • “Summarize this”
4. Press Ctrl+Shift+K again (or release the key)

   Recording stops. The selected text and your spoken instruction are both sent to the LLM.

5. Selected text is replaced

   The original selection is replaced with the transformed output. The clipboard and paste simulation happen in automation.rs using arboard and enigo — the same mechanism as regular dictation.

Example Transformations

Instruction	What the LLM does
”Make it professional”	Rewrites the text in formal, polished language
”Translate to Spanish”	Translates the full selection to Spanish
”Convert to bullet points”	Reformats prose as a bullet list
”Fix the grammar”	Corrects spelling, punctuation, and grammar errors
”Summarize this”	Condenses the selection into a shorter summary
”Make it shorter”	Removes unnecessary words while preserving meaning
”Add more detail”	Expands the text with relevant additional context
”Change to past tense”	Rewrites verbs in the past tense

How It Works Technically

When you trigger the hotkey with text selected, the Rust backend runs two operations before the overlay appears:

1. copy_selected_text() — simulates Ctrl+C via enigo to copy whatever is currently highlighted to the clipboard, then reads the clipboard value using arboard
2. get_active_window_info() — captures the foreground app for profile matching

Both the copied selection and your spoken instruction are forwarded to the frontend. dictationController.ts calls GroqChatService.transformText() with a prompt that combines:

• The original selected text
• Your spoken transformation instruction

The LLM returns only the transformed version of the text — no explanation, no preamble.

Original selection:  "hey can u send me the file when ur done thanks"
Instruction spoken:  "Make it professional"
Output pasted:       "Hi, could you please send me the file once you're done? Thank you."

Tips

You can chain transformations: paste the result, select it again, and apply another instruction. For example, first “translate to Spanish”, then “make it formal”.

Select-to-Transform works in any app that supports standard clipboard operations — code editors, email clients, browsers, word processors, chat apps, and more.

Select-to-Transform requires LLM post-processing to be enabled. If you haven’t configured an LLM API key yet, go to Settings > LLM Post-Processing and add your key.