Analyze & understand

Explaining via voice (Voice)

Simply dictate your case – ideal for complicated facts of the case while on the go.

Updated 27 October 2026

Some facts of the case are easier to tell than to type. Via voice input you describe your matter verbally in a conversation with Lex or Lexi, and the AI turns it into a structured basis.

How to proceed

Open the conversation.
Select your persona (Lex or Lexi) and – if you like – a conversation style (to the point, balanced, or listening).
Check the option "Lexi may look up information in my documents" (see below), tap on "Start conversation" and allow microphone access. Before the first conversation, you confirm a short notice once. This confirmation applies to your account on all devices; on a shared device, each person confirms it themselves. If the notice has changed – most recently because Lexi or Lex can now look up information in your account during the conversation (see below) – we will ask you to confirm it again. If you leave the page while the conversation is being set up, it will not start – already deducted credits will be refunded.
Explain your case in your own words. The AI summarizes and asks follow-up questions – just like with the facts of the case.
End the conversation using the corresponding button. Shortly after, the transcript and summary will be available in the history.

During the conversation

A countdown shows the remaining time; every conversation has a maximum duration.
Subtitles show what has been understood (can be shown/hidden).
Use a volume control and an "Enable sound" notice to ensure you can hear the AI – especially helpful on mobile.
The voice sounds more natural and responds faster than before.
If the character is currently looking something up in your account, "Lexi is looking through your documents..." (or Lex's name) will appear at the top. The short pause is not a glitch.

Speaking about a document, a legal question, a case file, or a proceeding

From the analysis of a document in My documents, an answered legal question, a case file, or an ongoing proceeding, you can "speak about it" – the conversation then knows the context. After the conversation, the recording can be assigned to a case file. The conversation mentions deadlines from your deadline register – with the checked or corrected date; it marks estimated or unverified deadlines as non-binding.

Looking up information in your account

During the conversation, Lexi (or Lex) can look up what is in your account instead of guessing:

Your matters (Organizing matters) and the facts recognized in relation to them,
Your uploaded documents – she only mentions a summary, never the full text,
Your open deadlines from the deadline register – only stored data, never calculated by her, and with the note to check the date in the document itself,
Your previous conversations.

If the account knows your name, she will address you by it. She speaks in your chosen language (German, English, French, Italian, Spanish, Dutch, or Polish).

You decide before every conversation: On the voice page, there is the option "Lexi may look up information in my documents" (with Lex's name for Lex). It is switched on by default, your device remembers your choice, and it is saved with the respective conversation. If the option is switched off, the character will look up nothing in this conversation – no matters, facts, documents, deadlines, or previous conversations.

Even with the option switched on, nothing is looked up for:

Documents that you have excluded from the AI, and deleted documents (trash),
Example and draft case files,
Lexus (short letters via voice) and the law firm assistant – they never look anything up.

Good to know: The AI masking (Protecting names and details from the AI) setting does not apply to voice – spoken language cannot be masked. What the character looks up and speaks is therefore not replaced by placeholders. If you do not want this, switch off the option before the conversation.

Information looked up is, like the rest of the conversation context, only transmitted to the voice service for this conversation, without AI training (Is it secure?). Because the scope has expanded, please confirm the notice again during your next conversation. If there is nothing yet, the character will say so honestly and invite you to upload a document to My documents.

Previous conversations in the text chat

If you ask in the chat about a previous conversation (e.g., "What did we discuss on the phone?"), the chat can read the summaries of your previous voice conversations. This is read-only, concerns only your own conversations and never the transcript; conversations under 30 seconds do not count. The chat mentions them as notes on the conversation, not as confirmed facts about your matter.

More than a general voice assistant

Unlike a general voice assistant, the conversation does not start from scratch: It knows your matter and your documents instead of having to be explained every time. General AI Chat or Legal AI? What Matters shows what this means in everyday life.

Summary via email

You can have a summary of your described case sent to you free of charge so that you have it in black and white.

Costs & extension

You can see the costs of the conversation before starting (Credits: The credit system, Where do I see what something costs before I start?). Optionally, you can activate an extension beyond the base duration – in that case, a per-minute price applies, which you also see in advance. Without checking this box, the conversation ends after the base duration at no further cost.

If a conversation with an extension ends unexpectedly (e.g., because the browser was closed), the additional minutes are billed based on the conversation duration measured by the voice service; if no measured duration is available, we do not charge anything extra.

Short letters via voice

From what has been spoken, you can directly generate short letters (Lexus, as PDF/DOCX via email) – see Short letters by voice (Lexus). Even after a conversation with Lexi or Lex, the summary offers to dictate a short letter or a small agreement with Lexus; this only begins after your click, the price is shown beforehand. If the letter mentions a deadline, you can adopt it as usual from its success page into your deadline register.

Tips for dictating

Speak calmly and in complete sentences.
State the topic first, then the details (Who? What? When? Which country?).
You do not need to formulate perfectly – the AI will ask follow-up questions if something is missing.

Your conversation history

You can find your previous conversations in the history on the conversation page. You can delete a conversation from the history once it is completed and billed – usually a few minutes after hanging up. If your conversation history could not be loaded, the page will show a notice with "Try again" – your conversations will not be lost.

If something does not work

No sound: tap "Enable sound" or check the volume.
Microphone denied: allow microphone permission in the browser/device and restart.
Connection drops: If a conversation is technically interrupted before it starts, the credits will be refunded (Refunds & Reimbursements).
"All lines busy": During very high simultaneous usage, a start may be briefly rejected – try again after a few minutes, no credits will be deducted.
Voice service temporarily disrupted: A notice banner on the page will indicate this before you start a conversation – please try again a little later.
Transcript missing initially: If the summary does not appear immediately after a conversation, it will be automatically generated in the background within a few minutes – no need to call again.

Afterwards, you can continue to Creating legal letters or to the legal question.

Dictation button in input fields

Regardless of the conversation, some input fields (e.g., for legal questions or in the writing assistant) offer a Dictate button. It uses the voice recognition of your browser: In Chrome and Edge, the recording is transmitted to the browser manufacturer's servers (Google or Microsoft) and converted into text there – SmartLegalPro only receives the recognized text. Before the first dictation on a device, you will see a short notice about this. Do not dictate anything that this provider should not receive, such as health data; typing is possible at any time.

Was this article helpful?