How it works
- TalkTap the mic and explain the idea like you would to a colleague. Fillers and false starts are fine.
- StopTap again. Your transcript is split into ideas and sorted into Context, Objectives, Tasks, Constraints and Questions.
- Fix & copyEdit the transcript if a word was misheard, hit Structure again, then copy the prompt into your AI.
Why dictate your prompts?
Most people speak three to four times faster than they type, and they include more context when talking. The problem is that a raw transcript is one endless sentence full of "um", "basically" and "and then", which AI assistants handle poorly: they latch onto the last thing you said and forget the constraint you mentioned at the start.
This tool keeps the speed of speech and adds the structure afterwards. It removes filler words in English and Spanish, uses spoken connectors ("also", "another thing", "y luego", "además") to split one long ramble into separate ideas, and then sorts each idea by intent. Phrases like "I want to…" become objectives, "add…" or "it has to…" become numbered tasks, "don't…" or "without…" become constraints, and anything that sounds like a question is kept as an open question, so the AI asks instead of guessing.
About privacy (the honest version)
Speech recognition in the browser is provided by the browser itself. Recent versions of Chrome can transcribe on-device for some languages; when that is available, the badge next to the language selector turns green and reads on-device. Otherwise, browsers send the audio to their vendor's speech service to transcribe it. CleanPrompt never receives your audio or your text, and the structuring step always runs locally. If that trade-off doesn't suit you, type or paste instead — the structurer works exactly the same.
FAQ
Which browsers support voice input?
Chrome, Edge and Safari on desktop and mobile. Firefox does not support the browser speech engine yet; you can still type or paste your notes and structure them.
Is my voice recorded or uploaded?
We never receive anything. Transcription is done by your browser's own speech engine: when the badge says on-device, audio stays on your computer; otherwise your browser may send audio to its vendor's speech service (Google, Microsoft or Apple). The structuring step always runs locally.
Can I dictate in Spanish?
Yes. Choose the dictation language before recording. The structurer understands English and Spanish and writes headings in either.
Do I need to speak in full sentences?
No. Filler words like "um", "basically" or "o sea" are removed, and spoken connectors such as "and then", "also" or "another thing" are used to split your ideas into separate items.
Why does recording stop by itself?
Some browsers end a session after a silence. The tool restarts it automatically while the mic is active; if it still stops, tap the mic again and keep talking. Your transcript is kept.