Contact
Real help from a real person.
Live Caption AI is a small team. Reply to any email we send, or write to us directly at hello@livecaptionai.com — Ryan reads them. Most answers are already below.
Fastest path
Email us.
We aim to reply the same day, and always within one business day. Include your account email, the platform you're using (iPhone, iPad, Android, or web), and — if it's about a specific session — the date and time it happened.
Frequently asked
Answers to what most people ask.
I'm not receiving my sign-in email.
Sign-in emails usually arrive within 30 seconds. Check your spam or junk folder first — Live Caption AI emails come from noreply@livecaptionai.com.
If it's not there, request another one; each request sends a fresh code.
If your email address has ever bounced (full mailbox, typo), we may have suppressed further sends — reply to hello@livecaptionai.com and we'll fix it.
The microphone isn't working.
On iPhone or iPad: open Settings → Live Caption AI → Microphone and switch it on.
On the web (Chrome, Safari, Edge): click the microphone icon in the browser's address bar and choose Allow, then reload the page.
On Android: Settings → Apps → Live Caption AI → Permissions → Microphone.
If the app crashes or freezes right after granting permission, quit and reopen it.
What's the best audio setup for a large room?
For small settings (meetings, classrooms, kitchen-table conversations), just place your phone within 6–10 feet of the speaker — the built-in microphone handles it well.
For churches, event halls, and conference rooms, plug your phone directly into the soundboard's auxiliary output using a TRRS adapter — the iRig 2 (~$40) and the Movo MC3 (~$10) are the two most common choices, and both work with any mixer that has a 3.5mm, RCA, or 1/4" aux out. This bypasses room noise entirely and gives you a pristine feed straight from the mic channel. Set the soundboard's aux level to around -12dB to -6dB — the app handles gain, and clipping hurts caption accuracy more than a slightly quiet signal. If your phone uses USB-C instead of a headphone jack (recent iPhones, iPads, and most Android devices), add a StarTech USB-C to 3.5mm TRRS adapter (~$15) between the phone and the TRRS cable.
For single-speaker events without a soundboard — a presenter walking a stage, a pastor at a lectern, a lecturer in a room without house PA — clip a dedicated wireless mic on the speaker and plug the receiver directly into your phone. The Rode Wireless GO II (~$300) is compact, battery-powered, and connects to your phone over TRRS or USB-C.
For hybrid events with multiple mics, mix them down to a single aux send before feeding the phone; don't try to switch mid-service.
— Product links above are Amazon affiliate links; we may earn a small commission at no cost to you.
How accurate are the captions?
Live Caption AI uses Deepgram Nova-3 for cloud captioning — currently one of the top-ranked models on public English speech-recognition benchmarks, with word error rates that consistently outperform earlier cloud services and open-source options like Whisper on real-world conversational audio.
End-to-end latency is under 500ms, so captions appear on screen effectively as fast as the words leave the speaker's mouth.
On-device captioning (Apple's SFSpeechRecognizer on iOS/iPad, Google's SpeechRecognizer on Android) is slightly less accurate but keeps your audio entirely private — it never leaves your device.
In practice, audio quality matters more than which recognizer you're using: a mic that captures the speaker clearly beats a better model with muddy input.
What are Specialized Models and Custom Terms — when should I use them?
Specialized Models tune the speech recognizer to specific vocabularies.
Medical knows pharmaceutical names, anatomy, and procedures (so "myocardial infarction" is transcribed correctly, not as the classic mishearing "mild car deal in farction"). Finance knows tickers, market terminology, and financial jargon. Worship handles scripture references, hymn titles, and liturgical vocabulary — spoken references like "John three sixteen" render as "John 3:16" automatically.
Custom Terms lets you add up to 100 of your own words or phrases — names of speakers in your congregation, your company's product names, technical acronyms — and the recognizer will prefer those spellings over similar-sounding alternatives.
Set these up before your event under Settings → Speech Recognition → Specialized Models & Terms. Doing this makes a bigger accuracy difference than almost any other tweak, especially for names and specialized vocabulary generic models tend to guess wrong.
How do I change or cancel my subscription?
Subscriptions are managed by whichever store you bought them through.
Purchased on iPhone or iPad: Settings → your name → Subscriptions.
Purchased on Android: Google Play → Menu → Payments & subscriptions → Subscriptions.
Purchased on the web: sign in, tap the Account tab, then Manage Subscription — this opens the Stripe billing portal where you can change plans, update your card, or cancel.
If you signed up on one platform and are trying to manage from another, the app will point you at the correct store.
Can I get a refund?
Refund handling depends on where you purchased.
App Store: request at reportaproblem.apple.com.
Google Play: use Play Store's refund policy in your order history.
Web (Stripe): email hello@livecaptionai.com within 14 days of the charge and we'll process it.
Legacy Bubble-era $3.99 subscribers who migrated to the new app: those receive a 50%-off Professional discount for 3 months automatically — no refund needed.
Broadcast or QR sharing isn't working for my audience.
Broadcasting requires a Professional or Team plan, or an active free trial.
Viewers scan the QR code from any phone's camera — the link opens in a browser, no app or account needed.
If viewers see a Session ended message, the broadcaster tapped End Session or lost their connection. If captions appear on the broadcaster's screen but not the viewer's, check that both devices are online — captions are streamed live, not stored.
Translation isn't showing up.
Live translation requires a Professional or Team plan and AI Data Processing consent granted.
You can toggle consent in the app at Settings → Privacy. Then on the Home screen, tap the Translate mode toggle before starting a session, pick your target language, and start.
If the target language you want isn't listed, our translation partner may not support it yet — reply to hello@livecaptionai.com and we'll tell you our rollout plan.
How accurate is the translation?
Translation is powered by DeepL, which consistently rates highly in independent blind translation studies — particularly for European and major world languages, where it often outperforms other services on nuance, idioms, and technical vocabulary.
Supported languages include Spanish, French, German, Italian, Portuguese, Dutch, Polish, Japanese, Chinese, Korean, Russian, and about 20 others; the full list appears in the app's translation picker.
Real-time translation lands within about one to two seconds of the original caption — fast enough to read comfortably in a live setting. DeepL uses neural machine translation with context awareness, so it handles a full sentence's meaning rather than translating word by word, which matters most for professional and religious settings where a small mistranslation can change the meaning of what was said.
If a language you need isn't in the picker, DeepL doesn't yet support it — let us know and we'll flag it in our roadmap.
AI summaries aren't generating.
AI summaries require a Professional or Team plan (or an active trial) and AI Data Processing consent granted.
They need at least about 30 seconds of transcript to have material to summarize. On the session detail screen, tap Generate Summary — if it fails, try again in a moment (the API can be busy).
Persistent failures are usually consent-related: open Settings → Privacy and confirm AI Data Processing is on.
Where's my data stored? Is it private?
By default, captioning runs on-device — your audio never leaves your phone.
When you enable cloud features (translation, AI summaries, high-accuracy captioning), audio is sent to our processing partners for real-time transcription and discarded — not stored.
Your transcripts are stored in your account, encrypted at rest, and are only accessible to you.
Full details are in our privacy policy at livecaptionai.com/privacy.
Try it