Dictation
Dictation is for speaking a draft the way you would dictate to a steno: at length, in your own language, with punctuation spoken aloud. Judicio transcribes as you talk, proposes corrections a paragraph behind you, and at the end offers to turn the transcript into a properly structured legal document.
Long-form dictation is rolled out account by account. Where it is not enabled the section simply is not on the Drafting home. There is no locked card, no upsell and no disabled button, and the server declines to start a session regardless of what the page shows.
This is not the same thing as the mic button on the composer. That is Voice input, which transcribes a sentence or two into a text field and is available to everyone.
Starting a dictation
The dictation section sits on the Drafting home, below the three-mode panel. It is headed "Speak your draft. Like dictating to your steno", with a line under it explaining how to shape the text as you talk: pause naturally between sentences, a longer pause starts a new paragraph, and spoken words like "comma" become punctuation.
Three notes sit beside it:
- The language is auto-detected. Hindi, Tamil, Telugu, Marathi and more.
- Corrections appear while you keep speaking.
- The formal draft can be in English or in the language you dictated.
| Control | What it does |
|---|---|
| Start dictating | Creates the session and opens it. Disabled when the browser cannot record, or while it is being created |
| Cost line beside it | States the base credits plus the per-minute rate, and that creating a formal draft afterwards is optional |
| Resume | Only when a previous sitting was interrupted. Names the draft and how much was recorded, and picks it up where it stopped |
There is nothing to configure before you start. There is no language picker, no microphone picker and no document-type field on this screen: the language is detected from your speech, and the choices about the finished document come later.
If your browser cannot record, a notice replaces the start block: Microphone not available, telling you to allow microphone access for this site or use a browser that supports recording. Saved dictations can still be resumed and drafted from.
The live session
Starting a dictation creates its drafting activity straight away, so the session lives at the ordinary draft address. The header carries the session title, or "Untitled dictation" until it has one.
Speaking, and being corrected
The transcript fills on the left as you talk. The rail on the right holds what the corrector found in the paragraph you have just left, so nothing interrupts you mid-sentence.


The header
| Control | What it shows or does |
|---|---|
| Status badge | Recording, Paused, Finishing or Stopped |
| Timer | Minutes and seconds of recorded audio. Paused time does not count |
| Live cost | The running credit estimate, refreshed from the server as you speak |
| Detected language | The language Judicio heard, marked as auto-detected |
| Output | Which language the finished draft should be in: English, or the language you dictated. It appears only once a non-English language has been detected, and takes effect immediately |
| Pause | Pauses recording. An icon control, named "Pause dictation" |
| Resume | Starts recording again after a pause |
| Finish | Ends the sitting and moves to the finish screen. There is no separate stop control |
| Discard | Abandons the sitting. It takes effect at once, with no confirmation |
The transcript
The panel is headed Live transcript, with the instruction "Click any line to correct it while you continue dictating." Before you speak it reads "Start speaking: your words appear here a moment after each pause."
Paragraphs are numbered in the gutter. A clip still being processed shows Transcribing…; one whose upload failed shows that it did not upload, with a Retry beside it. A paragraph the corrector has been through is marked as checked. A level meter animates while you are recording, and flattens when you are not.
Click any line to correct it by hand. It becomes a two-line editor, named "Correct this line": Enter commits, Escape restores what was there, and clicking away commits. Correcting a line by hand clears the suggestions for that paragraph, because they were worked out against the words you just replaced.
A footer reminds you of the two shortcuts: how long a pause starts a new paragraph, and the spoken words for a full stop and a new paragraph.
Corrections
Judicio checks each paragraph once you have moved on from it, so the checking never interrupts the speaking. Suggestions appear in two places at once, and resolving one resolves both.
- In the transcript. The misheard span is underlined. Clicking it opens a small panel showing what was heard and what is suggested, with Accept and Dismiss.
- In the Corrections rail. It is headed with the note that it is checked a paragraph behind you. Before anything is checked it reads "Nothing to correct yet. Suggestions appear once you finish a paragraph." Each card shows the paragraph number, the original and the suggestion, with Accept and Dismiss. Apply all takes a whole paragraph's suggestions at once.
Each suggestion is badged with its kind: Mishear, Punctuation or Grammar.
While you are paused, and when a limit is reached
Pausing shows a banner saying that nothing is recorded and nothing is charged while paused, and that your place is kept.
If your credit balance runs out mid-sitting, recording stops and a banner says so, along with the reassurance that everything so far is saved. The transcript footer then offers Finish now.
Resuming after a connection problem re-sends any audio that never reached the server, and tells you how many clips it recovered. If some cannot be recovered, finishing warns you plainly that those clips will be left out of the transcript for good, and offers to retry or to Finish without them.
Finishing
Finish stops the recorder, waits for the last audio to land, and closes the sitting. You then land on Dictation complete, with the length, the paragraph and word counts, the language, and the credits charged. A line under them says the transcript is saved to your activity and you can come back and choose later.
What the sitting leaves you with
The sitting is settled by the time you reach this screen. What you do with the transcript is a separate decision, and two of the three choices cost nothing.


Three peer choices follow. None is pre-selected.
| Choice | What it does | Cost |
|---|---|---|
| Create formal draft | Turns the transcript into a properly structured legal document | An ordinary standard drafting run |
| Open in editor as-is | Puts the raw transcript into the normal drafting editor as version 1, to edit, export or formalise later | Free. No AI touches it |
| Copy transcript | Copies the text to your clipboard. It stays saved either way | Free |
Leaving the page instead is safe. The run keeps the status Transcript ready in your history, and opening it brings you back to this same choice.
Turning a transcript into a formal draft
Create formal draft opens Create a draft from this dictation, which begins by saying that the transcript stays saved either way and you can re-create it with different settings later.
| Field | What it does |
|---|---|
| Document type | Free text, and optional. Name the instrument, or leave it empty and Judicio infers it from what you said |
| Draft language | English, or the language you dictated. Shown only when the dictation was not in English |
| Tone | Formal, Concise or Conversational. Defaults to Formal |
The footer states the cost as a standard drafting run with an estimate, and carries Cancel and Create draft.
There is nothing here about jurisdiction, files or templates. The session already holds those.
Creating the draft opens the ordinary drafting editor and streams the formal document into it. From there it behaves like any other draft: revise it with the assistant, save checkpoints, compare versions, export it. See Editing with AI.
If you opened the transcript as-is and later change your mind, the editor's header carries Formalise with AI, which opens the same dialogue. Its tooltip says what it does: rewrite the dictated transcript as a formal draft.
What a dictation costs
Pricing has two parts, and they are separate on purpose.
- The sitting is billed on recorded audio: a small base charge plus a per-minute rate, settled once when you finish. The header shows a running estimate as you speak, and the finish screen states what was charged.
- Turning the transcript into a formal draft is an ordinary drafting charge, and only happens if you ask for it. Opening the transcript as-is, and copying it, are free.
Exact figures are in Credits explained.
Finding a dictation again
Dictations sit in the same history as every other draft, marked with a Dictation badge and the note that they were dictated rather than drafted from files. Their status says where each one stands: Recording, Interrupted · Resume, Transcript ready, Completed or Discarded. Opening a row that is still recording, interrupted or waiting on your choice takes you back to the dictation surface; a completed one opens the editor.
Troubleshooting
The microphone is blocked
The banner reads that microphone access is blocked and asks you to allow the microphone for this site, then try again. Allow it in your browser's site settings and click Try again.
No microphone was found
Connect one and click Try again. If another application is holding the microphone, close it and try again.
Voice dictation is not available in this browser
Recording needs a browser that supports it. You can still resume a saved dictation and create a draft from it.
A clip did not upload
The line shows that the clip did not upload, with a Retry. Retry before finishing: finishing while clips are unrecovered leaves them out of the transcript permanently, and Judicio says so before it lets you.
Not enough credits to start a dictation
Top up or change plan, then start again. See Managing your subscription.