Frequently asked
questions

Buzz supports 40+ languages across four STT providers. Western languages (English, Spanish, French, German, Japanese, Chinese, Korean, Arabic, and more) are powered by Deepgram. Indic languages (Hindi, Bengali, Tamil, Telugu, Kannada, Malayalam, Marathi, Gujarati, Punjabi, and more) are powered by Sarvam AI. Some additional languages use Gladia. See the full language list →

Yes. The Free plan is restricted to English (US, India, UK, Australia). Basic and Pro plans support all 40+ available languages.

Translation is available on Basic and Pro plans. Basic supports translation to English only. Pro supports translation to any available target language, including English, Spanish, French, German, Japanese, Chinese, Korean, Arabic, Portuguese, Hindi, and all Indic languages via Sarvam. Translation is not available on the Free plan.

For Indic languages like Hindi, Tamil, Bengali, and others, you can choose between native script output (e.g., हिन्दी) or Roman/Latin transliteration (e.g., "Mera naam Rohit hai"). Transliteration makes the text readable without knowing the native script. Both modes use Sarvam's saaras:v3 model.

Accuracy varies by language, accent, audio quality, and background noise. Under ideal conditions (clear speech, low background noise), Deepgram Nova-3 achieves very high accuracy for English and major Western languages. Sarvam performs well for Indic languages. Results are best effort — Buzz is not suitable for legally binding, medical, or safety-critical transcription. You can always edit any segment inline.

Yes. You can select multi-language modes that handle sessions where speakers switch between languages. Gladia supports up to 10 language hints for code-switching. Deepgram and Google also offer multi-language detection modes. Select "Multi-language" from the language picker when starting a session.

Buzz supports three session types: Solo (one person, live microphone), Group (up to 10 participants, each with their own audio stream, joined via QR code or 6-character code), and Voice Note (upload a pre-recorded audio file for batch transcription).

A single session can last up to 4 hours (240 minutes). This is a hard cap regardless of your plan. Sessions auto-end when this limit is reached. For longer events, you can start a new session.

Buzz supports MP3, M4A, AAC, OGG (WhatsApp and Telegram), WAV, WebM, FLAC, CAF (iMessage/iOS), and MP4. The maximum file size is 50 MB. Uploaded audio is routed to the best provider for the selected language: Deepgram (non-Indic languages, with speaker diarization) or Sarvam (Indic languages like Hindi, Tamil, Bengali, etc.).

Yes. You can tap any transcript segment in the session history to edit it inline. You can fix words, correct speaker names, or delete segments. Edits are saved server-side and persist across devices.

Speaker diarization is available for all languages processed by Deepgram (Western languages: English, Spanish, French, German, Japanese, etc.). It is not available for Indic languages processed by Sarvam AI. For uploaded Indic-language voice notes (routed via Sarvam), speaker diarization is not supported.

No. Buzz requires an active internet connection for all features. Transcription, translation, session history, and account management all require connectivity. There is no offline mode.

Buzz is available for iPhone (iOS 16+). The Android app (Android 7.0+) is currently in closed testing — public release coming soon. The app is not available for desktop or web browsers.

Export and sharing requires the Pro plan. Pro users can share as plain text, export as a formatted PDF with speaker labels, timestamps, word count, and multi-script font support. Free and Basic plans have no export or sharing functionality.

Yes. Buzz uses Google's Noto Sans font family for PDF export, which covers Devanagari, Tamil, Telugu, Kannada, Malayalam, Bengali, Gujarati, Gurmukhi, Arabic, Thai, Sinhala, Khmer, Myanmar, Georgian, Hebrew, CJK (Chinese/Japanese/Korean), and Latin scripts.

Export and sharing is Pro-only. On the Pro plan, use the share button to share as plain text via any app (WhatsApp, email, Messages, etc.), or export as a PDF. There is no public session link or URL sharing.

No. Audio is never written to disk or stored anywhere. It passes through server memory only — streamed to the STT provider and discarded immediately when the session ends. All in-memory audio buffers are cleared at session close.

Yes. Transcript text is encrypted with AES-256 before being stored in Firestore. The encryption key is held server-side and never sent to the app.

Session retention depends on your plan: Free keeps sessions for 7 days, Basic for 30 days, and Pro for 90 days. Sessions older than your limit are automatically and permanently deleted, including all transcript segments, speaker labels, and associated exports. You can also delete any session manually at any time.

Buzz does not operate any proprietary AI models, so we have no means to train on your data ourselves.

For the third-party providers that process your audio and transcripts (Deepgram, Sarvam, Gladia, Google, DeepL), we are actively working to confirm and enforce data-processing agreements that prohibit your data from being used for model training. This is an ongoing process — we will update this page as agreements are finalised. In the meantime, refer to each provider's privacy policy for their current stance on training data use.

See the Privacy & Trust page and our Privacy Policy for full details.

You can request account deletion via the Delete Account page or by going to Settings → Profile → Delete Account in the app. We'll process your request within 7 days and delete all associated sessions, transcripts, usage data, and your Firebase Auth record.

You won't be able to start new sessions until your quota resets at midnight (in your local timezone). If you have a credit pack, credits are used automatically when your plan quota runs out — so you can keep going seamlessly. If you have no credits, you'll be prompted to purchase a pack or upgrade.

Credit packs are one-time purchases that add extra transcription time to your account. They work on any plan and are consumed after your plan's included quota runs out. Available packs: 15 minutes, 60 minutes, 120 minutes, and 300 minutes. Credits expire 60 days from purchase and are non-refundable.

Cancel through your App Store (iOS) or Google Play (Android) account settings. You'll retain access to your current plan until the end of the billing period, then automatically return to the Free plan. We don't handle billing directly — all subscriptions are managed by Apple or Google.

Refund eligibility is determined by Apple or Google, not by us. For subscriptions, submit a refund request through the App Store or Google Play. Credit packs are non-refundable once purchased.

Credits are consumed during a session when your plan quota runs out — not at the start. If you've already used your daily/monthly quota and have credits remaining, you can still start a new session. If you see a "limit reached" message, check that your credits haven't expired (60-day expiry).