The sun rose over the city. People walked to work. It was a normal day.
How Voice Editing Works
Voice editing is the heart of Fable. Instead of typing commands or clicking through menus, you press record, say what you want in plain language, and the AI makes it happen.
Press record
Speak your instruction
Fable transcribes locally
The AI edits your document
History is saved
What You Can Say
You can give any kind of instruction — from small tweaks to sweeping changes:
- "Make the opening paragraph more dramatic"
- "Add a scene where the characters arrive at the castle"
- "Fix the dialogue in the third paragraph — it feels stiff"
- "Change the tone to be more mysterious"
- "Remove the last two paragraphs"
- "Write the opening scene of a mystery novel set in 1920s Paris"
The AI can edit existing text and generate new content from scratch. Open a blank document and describe what you want — the AI will write it for you. But the real magic is in editing and refining your own writing.
The Recording Experience
While recording, an animated indicator gives you real-time visual feedback that Fable is listening. The record bar also displays your last instruction for context, so you always know what you asked for most recently.
After you stop recording, you'll see a brief processing state while Fable transcribes your voice and sends the instruction to the AI. Then changes start streaming into your document — you can watch paragraphs appear and transform in real time.
A Respectful AI Editor
Fable's AI is designed to be a careful editor, not a ghostwriter:
- Preserves your voice — it matches your style, tone, and word choices
- Makes targeted changes — it only touches what you asked about
- Never rewrites unless asked — it won't restructure your whole document uninvited
- Always explains itself — every edit comes with a clear explanation
- Works with any kind of writing — fiction, non-fiction, scripts, notes, outlines, poetry
Privacy
Your audio never leaves your machine. Fable transcribes your voice locally using built-in speech recognition (Whisper). Only the transcribed text and your document content are sent to the server for AI processing.
When you give a voice instruction, the AI only sees the current document and what you asked for. It doesn't have access to your other documents or projects.