Frequently asked questions

Video-call and in-person speech audio is sent to Deepgram for real-time speech-to-text. Each complete utterance is then sent to Google Cloud Translation in the listener's selected language. A downloaded on-device speech model may be used as a backup if cloud transcription fails.

Both the app interface and calls are available in 41 languages. Availability and translated-voice support can vary by device.

No live translation mode currently works offline. Speech recognition, translation, and delivery require internet; a downloaded on-device speech model is only a transcription backup.

AI Voice is an optional feature that speaks your translated words in a voice that sounds like you. You record a short sample once, and from then on the people you talk to hear your translated words in your AI voice instead of a standard one, on video calls, voice calls and in person. It is still a synthetic voice, and listeners see an “AI Voice” label while it speaks. There is more on the AI Voice page.

No. AI Voice is included at no extra cost. A call with it uses minutes exactly as any other call does, and only the host spends minutes.

In the app, record yourself speaking for 30 to 60 seconds. During the recording, the app shows you four words to read aloud; that confirms the recording is being made live, by you, and is not a playback. You can only create an AI voice from your own voice.

You can set it up in 40 of the 41 languages (not yet Hebrew). People listening in Bengali, Gujarati, Hebrew, Kannada, Marathi, Serbian, Thai, Traditional Chinese or Urdu hear a standard voice.

Yes. Whenever your AI voice is speaking, listeners see an “AI Voice” label, in the app and on the guest page. If anyone listening is using a version of the app too old to show the label, everyone in that conversation hears your standard voice, not just that person.

A guest who joins from a link in a browser cannot create an AI voice, but hears yours, with the label. Anyone who joins with the app can use their own AI voice, in sessions they host and in ones they join.

In most countries. It is not offered in China, Malaysia or South Korea, in countries and regions under comprehensive US sanctions, or in the US states of Illinois, Texas, Washington and Colorado, which have their own laws on voice data. Where it is not offered, your translations are spoken in a standard voice and everything else works the same.

VivaLingua does not keep it. It passes through our servers to Deepgram, which checks the four words, and to ElevenLabs, which makes your AI voice from it; your name and email address are not sent with it. ElevenLabs keeps the voice model until it is deleted, and does not use your recording or your voice model to train or improve its AI. The privacy policy treats both as biometric data and has the full detail.

In Profile, any time. It stops being used at once, and we ask ElevenLabs to delete it straight away. It is also deleted automatically if you do not use it for 12 months or do not use VivaLingua at all for 6 months, and when you delete your account.

Buy prepaid minute packs when you need them. There is no subscription, and purchased minutes never expire.

30 minutes for $2.99, 60 minutes for $4.99, or 120 minutes for $5.99 — normally $8.49, on offer until October 31. Purchased minutes never expire.

It depends on the mode. On a call, video or Audio Only, minutes run from the moment the call connects until it ends — silence counts, because the call is open the whole time — and time in the waiting room before it is free. In an in-person session, time counts while someone is speaking or being translated, and for up to 45 seconds after; longer quiet stretches do not count. Either way it is only ever the host's balance.

No. The number of guests makes no difference — only the host spends minutes, and in an in-person session what is counted is the translations, not how many people are in the room.

VivaLingua warns you as the balance runs low, and the session ends when it reaches zero. Buy another pack and you can start again right away. Purchased minutes never expire, so whatever is left over is still there next time.

Yes, you get 60 promotional minutes when you sign up: straight away if you use Google or Apple sign-in, or once you confirm your email address if you sign up with email. They are given once per person: a new account does not get them again. Promotional minutes carry no cash value and expire 90 days after they are granted, and they are spent before any minutes you have bought — so a pack you paid for is never used up while free ones are still sitting there.

Only the host uses minutes. Guests always join calls and in-person sessions free.

Yes. Anyone can host a call or in-person session using minutes from their own account — including right after being someone's guest. You can call them straight back.

Email contact@vivalingua.app. Purchases are processed through the App Store or Google Play depending on how you bought them, and refunds follow that store's own policy; amounts already paid are otherwise non-refundable except where the law requires it.

Yes. Tap “Start Audio Only call” on Home, or the phone button beside a contact. No camera is used on either side, and the screen shows the conversation as text: what each of you said, in both languages. Translation, languages and minutes are the same as on a video call, and it uses less data. A guest you invite by link can join an Audio Only call too.

You hear it. VivaLingua speaks the translation out loud in the other person's language and shows captions on screen at the same time — you can adjust both in the in-call settings. If the speaker has set up AI Voice, it is spoken in a voice that sounds like them. That is the difference from tools that only caption: you can look at the person instead of reading subtitles while they wait.

VivaLingua translates one sentence at a time. When you finish a sentence and pause, the other person hears it in their language, usually about 2 to 3 seconds later, with captions on screen at the same moment. Some language pairs are a bit slower than others.

Speaking in full sentences with a short pause between them gives the smoothest conversation. A weak or mobile-data connection on either side can add a little delay, so Wi-Fi on both ends helps. On a weak connection, an Audio Only call uses the least data.

Both translate, in narrower ways. FaceTime shows translated captions on one-to-one calls — you read the translation rather than hear it — and it needs Apple Intelligence, so recent Apple devices only, across roughly ten languages. Google Meet does both: its translated captions cover around 70 languages, but the version that actually speaks covers only English paired with five others (Spanish, French, German, Portuguese and Italian), on a paid Google plan. VivaLingua speaks and captions, across 41 languages that translate directly to one another rather than through English, on both iPhone and Android. With AI Voice, it can speak your translated words in a voice that sounds like you, in most of those languages, at no extra cost. So an Arabic speaker and a Thai speaker can hear each other, which neither of the above can do. It also works face to face, not just on calls, and only the host spends minutes — your guest pays nothing and needs no subscription. There is a fuller side-by-side comparison if you want the detail. Checked September 2026; both are changing quickly.

Yes. Calls require an internet connection.

Wi-Fi is recommended. When Wi-Fi is unavailable, VivaLingua uses your mobile data connection.

Open in-call settings to control translation captions, view the session voice, and adjust translated and original audio.

If the person you call is using VivaLingua right now, their phone rings as usual. If they are not, a waiting room opens instead and they get a notification inviting them in. The same happens if they miss or decline the call. If they have an older version of the app, their phone rings as usual.

You can both see who is in the room. When the other person turns on “Available to talk now”, you can tap “Start call”. If you turn it on too, the call starts by itself as soon as you are both there. A guest you invite by link waits in the same room, in their browser. Waiting there is free, and the room closes after 30 minutes.

If you started the call and the other person’s connection drops, you can end the call or wait for them. Waiting takes you back to the waiting room, where the call can start again when they return. If you do nothing for 30 seconds, the waiting room opens by itself. Time in the waiting room is free.

Yes. In-person speech audio is sent to Deepgram for speech-to-text, and finalized text is sent to Google Cloud Translation. A downloaded on-device speech model may serve as a backup, but live translation still requires internet.

Choose Push-to-Talk to hold the microphone button while speaking and release it when finished, or choose Hands-Free to speak naturally while VivaLingua detects each turn.

In-person speech audio is processed by Deepgram for real-time speech-to-text. Finalized text is sent to Google Cloud Translation and relayed for delivery. VivaLingua does not persist ordinary source transcripts; see the Privacy Policy for how these providers handle your data.

No. You need it to host, but anyone you invite can join from a link in any browser — no app, no account, nothing to sign up for. They get what an app user gets: they hear you in their language and you hear them in yours, spoken out loud. A video call takes one guest this way; in an in-person session you can let in more than one, in any mix of people using the app and people on a link. Either way, guests pay nothing.

Tap “Invite by link” and your phone’s normal share sheet opens — the same one you get when sharing a photo. For a call, the waiting room opens first and the share sheet appears by itself. Send it by text, WhatsApp, email, Telegram, Messenger or Signal, or copy the link and paste it wherever you like. VivaLingua does not send the invitation itself; it hands the link to whichever app you already use. An in-person session also offers “Show QR”, for someone at the same table who would rather scan than wait for a message. Video calls have no QR code, because the other person is usually not in the room.

The message reads “Talk with me in your own language. Open this and I will let you in”, in whatever language your app is set to; apps that use an email subject get “Join my VivaLingua conversation”. Your guest opens the link in any browser and chooses their language. For a call, they wait with you in the waiting room and turn on “Available to talk now”, and then you tap “Start call”. For an in-person session, you tap “Let them in”. There is nothing for them to install and no account to create.

It stays the same. You keep one link, so sharing it twice sends the same one. It only works while you are actually in a session, though — sending it is not enough on its own, which is why the app reminds you to start the session once you have shared it.

Standard Bluetooth earbuds and the phone speaker work with VivaLingua.

Video-call and in-person speech audio is processed by cloud speech-to-text providers, and finalized text is sent for cloud translation. VivaLingua does not persist ordinary source transcripts and temporarily caches translated results for reliable delivery as described in the Privacy Policy.