Help centre

Any Windows application with a text field. It types into whichever field has focus, the same way your keyboard does — email clients, chat apps, browser forms, code editors, ticket trackers, the search box. There is nothing to install per application.

No, the hotkey is configurable. You can also choose between holding the key while you talk and pressing it once to start and again to stop.

Only if you ask it to. Launch at startup is an option, off by default. When it is running, a clear indicator shows whenever the microphone is live, and it disappears the moment recording stops.

Yes. That is the case Xuula is built around. You do not select a language beforehand or change your keyboard layout — you speak, and the recognition follows the switch. Russian and English are the launch focus, and your account can store several preferred languages.

Accuracy depends on your microphone, your accent, background noise and how clearly you speak. We do not publish an accuracy figure, because a single number would not tell you anything useful about your own setup. The free plan exists so you can find out on your own hardware.

Yes. Turn cleanup off in settings and the text is inserted as recognised, with no editing pass. When cleanup is off, no transcript text is sent to the rewriting provider at all.

Names, brands and technical terms that a general model gets wrong — colleagues’ surnames, your product names, your stack. You add the correct spelling and, optionally, the variants it is usually misheard as. It is a Pro feature.

It does not join anything. Xuula records locally on your PC: your microphone and, when enabled, the Windows system audio — the sound of the call you can already hear. That works with any app that plays sound, with no meeting-platform account or bot. It cannot capture audio Windows will not expose, such as DRM-protected or some exclusive-mode audio paths.

On your PC, as ordinary local files: the recording, the final transcript, the translation, your notes, speaker names, word timestamps and the AI insights, organised into Projects. Nothing from your meetings is uploaded to Xuula servers in this release, and there is no cloud sync. You can back the files up or delete them yourself.

It separates speakers acoustically and labels them, and you can rename yourself, the meeting audio and each detected speaker for the session. That is separation, not identification — it does not verify anyone’s identity, and renamed labels are simply names you assigned.

Yes, optionally. The original transcript and a translation into a target language you choose are shown side by side — in the workspace and in movable caption windows. The original is always kept; translation can be turned on or off per meeting.

They are drafts generated from the transcript, on your request only — nothing runs in the background. AI output can miss context, mangle a name or misread intent, so review it before acting on it. Action items are text for you to use; they are not sent anywhere, assigned to anyone or synced to a task manager.

That depends on where you and the other participants are: some jurisdictions require everyone’s consent, others only yours, and workplace or contractual rules can add more. You are responsible for giving any legally required notice and obtaining consent before recording. Xuula does not make that determination for you.

No. Speech recognition runs in the cloud, so an internet connection is required. Hotkey handling, the recording indicator, text insertion and your local history all run on your PC, but the recognition itself does not.

Xuula servers do not store your audio or your transcripts. Audio is streamed to a speech recognition provider while you are recording. That provider has its own retention terms, which are stated on the Privacy Policy page only where they have been confirmed — where they have not been, the page says so rather than making a claim we cannot back.

No, and it is designed never to see one. Account sign-in for the desktop app works through your browser: you approve the device there, and the app holds only a short-lived token it can refresh. You can revoke any device from your account settings, which takes effect within minutes. Any provider keys the app itself is configured with are kept in Windows Credential Manager, not in plain files.

Yes, both from your account settings. Export produces a JSON file with your account details, usage ledger, dictionary and styles. Deletion removes your data after a short recovery window, except records we are legally required to keep, such as invoices.

Two thousand words every week, with no credit card. That includes mixed-language dictation, use in any Windows text field, basic cleanup and local dictation history. The allowance resets weekly on a fixed schedule; your account page shows the exact reset time in your own time zone.

The annual plan is $120, charged once per year. That works out at $10 a month, which is how it compares with the $12 monthly plan — but the charge itself is a single $120 payment, not a monthly one. You save $24 over a year.

Two authorised devices on the free plan and five on Pro. You can see and revoke authorised devices from your account settings at any time. The plans are for personal use, not for sharing an account across a team.

No weekly word cap. There is a published fair-use threshold so one runaway account cannot degrade the service: we get in touch if an account passes 500,000 words in a month and give you a week to respond before any limit applies. A single dictation is capped at 12,000 words. The exact wording is in the Terms of Service.

Pro stays active until the end of the period you have already paid for, then the account returns to the free plan. Nothing is deleted — your dictionary, styles and settings are still there if you come back.

We offer a 14-day refund window on a first subscription payment, on top of whatever statutory rights apply where you live. The Refund Policy page has the details and the exceptions.

Check that the target window still had focus when you released the key, that Windows has microphone permission for Xuula, and that the recording indicator appeared at all. Applications running as administrator can also refuse input from a non-elevated app — running Xuula with the same privilege level fixes that.

Check that system-audio capture is enabled for the meeting and that the call audio is playing through the device Windows is using — audio routed to an excluded or exclusive-mode device may not be capturable. DRM-protected audio paths cannot be captured at all. Your own microphone also has a mute switch on the mini panel; make sure it is not the other way around.

Add the word to your personal dictionary with the spelling you want. That is exactly what the dictionary is for: proper nouns and technical terms that sound like a word in the other language.

Dictation stops until the allowance resets, and the app tells you when that will be. Your account’s usage page shows the exact reset time in your time zone. Upgrading to Pro removes the weekly cap immediately.