Find a quick fix

Press Cancel translation, confirm that you are online and retry once. A new uncached entry can take longer because Qelumi must validate the source language and build an authoritative translation list. If every word freezes, check the backend health page. Send support the exact word, language pair and time if it continues.

This normally means the browser/app could not reach the secure backend. Check internet access, disable a restrictive VPN/ad blocker temporarily, make sure the date/time on the device is correct and retry. If the health page also fails, contact support.

Some words belong to more than one language—such as “apocalypse” or “saboteur”—and short entries have little evidence. Keep the selected language when the entry is valid there. For a clearly wrong direction, accept the swap. Include the exact word and pair in feedback if Qelumi rejects a valid use.

Qelumi asks whether you want to change one language. If you continue, it gives a monolingual definition and source-language details without duplicating destination content.

This is expected for a completely new entry: the complete translation list is intentionally delivered first, while word origin, meanings, conjugations, phrases and examples are prepared separately. Keep the app open. If the button reports an error, retry the same search; complete cached data should load faster afterward.

Open the in-app feedback form, choose Translation Error and include the source entry, language pair, missing translation/sense and your expected correction. AI-generated language content can be imperfect.

Confirm that More details finished. Conjugations appear only for entries classified as verbs, and phrase entries may not have a conjugation. Examples are grouped by accepted meanings. Report the exact missing tense or meaning if the generated entry remains incomplete.

Press Enter to submit and close the list, or tap outside the search field. Autocomplete depends on network access but translation does not require choosing a suggestion.

Open device settings, find Qelumi or your browser, and allow Microphone access. On iPhone/iPad also check Settings → Privacy & Security → Microphone; on Android check Settings → Apps → Qelumi/Browser → Permissions. Reload Qelumi after changing permission.

Wait until the visible listening message appears, speak one clear word or sentence near the microphone and reduce background noise. Confirm the exercise language matches your speech. Pronunciation Coach allows repeated attempts.

Qelumi reconciles interim and final browser speech results. If repetition returns, press Stop, use Clear and begin a new session. Avoid switching apps while recording. Send the device/browser and a screenshot or short example to support because speech engines return different result formats.

Keep the Qelumi window active and the screen awake. Some mobile browsers impose speech-session limits; Qelumi restarts supported sessions, but the operating system can still stop listening when the app loses focus.

Use Qelumi’s built-in recording button where possible. If you upload audio, use a common format such as M4A, MP4 audio, WebM, WAV or MP3. Very large files or unusual codecs may be rejected.

Translate more entries first, then check Study Focus. A custom list, Weak Words or Recall List with no eligible words can leave a game empty; switch Study Focus to All History.

Qelumi can only use voices exposed by the device/browser. Install the matching speech voice in system accessibility or text-to-speech settings, then fully close and reopen Qelumi. If still unavailable, Qelumi names the default voice it will use.

Open Android Settings and search for Text-to-speech or Speech services. Open the preferred engine, install voice data for the desired language/region, select it once and restart Qelumi. Menu names vary by manufacturer.

Open Settings → Accessibility → Spoken Content → Voices, choose a language/region and download a voice. Also check Settings → General → Language & Region. Restart Qelumi afterward.

Check the source-language selection and device voice availability. Qelumi requests the relevant locale, but the operating system makes the final voice choice. Report the source language, chosen accent and the “default voice” message if it persists.

Raise media volume, disable silent mode where it affects speech, disconnect unwanted Bluetooth devices and test another speaker button. Some browsers require one user tap before allowing audio.

Allow Camera and Photos/Files access for Qelumi or its browser in device settings. On the web, the camera also requires HTTPS—which qelumi.com provides.

Use a sharp, well-lit image, keep the text horizontal, crop closely around the relevant lines and avoid glare or handwriting where possible. Select the correct source language before translating.

Use JPEG, PNG or WebP and crop to the text you need. A screenshot is usually smaller and clearer than a full-resolution camera photo.

Confirm the email spelling and use Forgot Password if the account exists. If you registered by Google or Apple, use the same provider instead of an unrelated password. Verify the registration email if Qelumi asks.

Check Spam/Junk and search for Qelumi or Firebase. Wait several minutes before requesting another email. Confirm you used the registered address and that its mailbox is not full.

The current Qelumi configuration uses qelumi.com as the authentication domain. If an old cached screen appears, close all Qelumi windows and reopen. The Firebase project identifier may still appear in technical URLs without changing the Qelumi service.

Make sure you signed into the same account/provider. Anonymous data belongs to that anonymous session and may not follow you to another device. Stay online briefly so Firestore can synchronise.

Supported Save controls change to Saving… and then a saved state. If it remains busy, check your connection and sign-in. Open the matching History, Review or Lab folder to confirm.

Sign in with a registered account. Use 3–24 permitted characters and choose a unique name. Reserved words such as “admin”, “support” and “qelumi” cannot be claimed.

That is an intentional security check. Sign out, sign back into the exact account and retry immediately. You can also use qelumi.com/delete-account.

Some games require at least five eligible words or saved context examples. Translate more enriched entries, save examples and set Study Focus to All History until you have enough data in a smaller list.

Open Learn → Study Focus and check whether All History, Weak Words, Recall List or a custom list is selected. The chosen focus applies to compatible games.

The badge reflects unrevised/due vocabulary. Complete reviews or use the visible Clear badge control to dismiss the current notification. Future due work can make it appear again.

Enable device reminders in Recall List and allow notifications in system settings. Battery optimisation, Focus/Do Not Disturb and browser restrictions can delay reminders. The recall schedule remains visible in Qelumi even without a notification.

The level primarily changes vocabulary, grammar complexity and response expectations. Qelumi can intentionally offer the same mission catalogue at several levels so you can practise the same situation with harder language.

Sign in and retry. AI-backed Lab features need an authenticated session so the secure backend can authorise and rate-limit requests.

Use the complete eight-character invite code before it expires, or verify the exact username. A room accepts only two participants. Both people must be signed into registered accounts and online.

Wait until the room shows both participants, keep both apps open, check each person’s selected language and confirm internet access. Leave and create a new room if the invite became stale.

Open Lab and select the matching saved folder: Context Lens, Live Transcripts, Text Translations, Conversations, Grammar Reviews, Writing Reviews or Read Aloud Texts. Confirm you are in the same account.

Speed is requested from the browser/device speech engine and can vary slightly. Regional accents require an installed matching voice; otherwise the default same-language voice is used.

Confirm that Firebase deployed the correct project and complete public/ folder. Close every Qelumi tab and installed-app window, wait a moment and reopen. The service worker uses a safe network-first/stale-while-refresh strategy, so a second reopen may be needed after an update is downloaded.

The app shell and some cached/local data can still open, but new AI translations, sign-in refresh, cloud saving and connected features need internet access. Restore the connection and reopen the requested feature.

Remove the installed web app/shortcut, clear only the site data for qelumi.com if necessary, revisit qelumi.com and install it again. Your cloud learning data remains tied to your account, but unsynchronised anonymous local data may be lost.

Speech recognition, installed voices, background activity, camera access and notifications are controlled partly by the operating system. Qelumi provides fallbacks where possible, but exact availability can differ by device and browser.