Privacy Policy
WA Transcriber
Effective date: 20 August 2026
Last updated: 20 August 2026
Binaryscript Private Limited ("we", "us") operates the Android application WA Transcriber (the "App"), published on Google Play as com.binaryscript.watranscriber.
This policy describes exactly what the App does and does not do with your data. It is written against the App's source code rather than in general terms, so where it says something never happens, that is a statement about how the App is built.
The short version
Voice notes are transcribed entirely on your phone. The audio you share with the App, and the text produced from it, are never uploaded to us or to anyone else. We operate no servers and have no account system, so there is nowhere for your recordings to go.
What does leave your device is limited to: a one-time download of the speech model, anonymous usage and crash statistics, and your purchase status through Google Play. Each is described below.
1. What stays on your device
| Data | Where it is stored | Leaves the device? |
|---|---|---|
| The audio you share | App cache, overwritten by the next one | No |
| The transcript text | files/history.json | No |
| Your speech model | files/models/ | No |
| Settings (language, model, translate) | App preferences | No |
| Free-usage counters | App preferences | Only via Android Auto Backup — see §5 |
| Purchase status cache | App preferences | No |
Speech recognition runs locally using whisper.cpp, an open-source engine bundled inside the App. It works with no network connection at all once the model has been downloaded.
We never see your audio or your transcripts. No copy is transmitted, analysed, used for training, or shared.
2. What the App sends, and why
2.1 Speech model download
The first time you transcribe, the App downloads a speech model (32–190 MB) from Hugging Face (huggingface.co), a third-party host. That request exposes your IP address to Hugging Face in the ordinary way any download does. No information about you, your device or your audio is attached to it. The download happens once per model.
2.2 Analytics (Google Firebase Analytics)
We collect anonymous usage statistics to understand how the App is used and which parts fail. The events are:
share_received, transcribe_started, transcribe_completed, transcribe_failed, model_download_started, model_download_completed, model_download_failed, free_quota_exhausted, paywall_viewed, begin_checkout, purchase, subscription_purchased, purchase_restored, purchase_cancelled, purchase_failed, trial_started, review_conditions_met, review_flow_launched, review_flow_completed, review_flow_failed, more_apps_page_viewed, more_apps_app_clicked, settings_link_clicked, update_clicked.
Alongside these we record three properties: whether you are a paying user (is_premium), which speech model you have selected (model), and the transcription language you have chosen (app_language).
No content is ever included. Audio, transcript text, file names and content URIs are excluded by design. Where a measurement would otherwise describe your recording, it is recorded as a coarse bucket instead — an audio clip is reported as "10–30s", never its length; a transcript as "50–200" characters, never its text.
Firebase Analytics assigns a per-installation identifier and collects standard device information (model, operating system version, country, language). This data is processed by Google; see Google's privacy policy at https://policies.google.com/privacy.
Our Firebase project is linked to Google Ads so that we can measure which advertising campaigns lead to installations and purchases. This means the events above, and the installation identifier, are shared with Google for analytics and advertising measurement.
2.3 Crash reporting (Google Firebase Crashlytics)
When the App crashes or hits an error, a report is sent to Crashlytics containing the stack trace and technical context: device model, ABI, Android API level, which speech model was in use, the number of audio samples being processed, the thread count, the CPU features reported by the speech engine, and which stage failed. Most crashes in this App happen inside native speech-engine code, and without that context a report is unusable.
Crash reporting is disabled in development builds and enabled in the version published on Google Play.
2.4 Remote configuration (Google Firebase Remote Config)
The App periodically fetches settings — the minimum supported version, the free usage limit — from Firebase. The request carries an installation identifier and basic device information. It sends nothing about your use of the App.
2.5 Purchases (Google Play Billing)
Purchases are handled entirely by Google Play. We never see your payment details. The App receives from Google Play whether you hold an active purchase, which product it is, and an opaque purchase token, and caches these on your device to unlock paid features. We hold no server-side record of your purchase. See also our Refund Policy.
3. What we never collect
- Your audio, in any form.
- Your transcripts, in whole or in part.
- Names, email addresses, phone numbers, contacts or your address book.
- Your location.
- The contents of any other app, including any messaging app.
The App requests no runtime permissions at all. It holds only the INTERNET permission, used for the four purposes in §2.
4. Legal bases and your rights
Where the EU/UK GDPR applies, we rely on legitimate interests (Article 6(1)(f)) for anonymous analytics and crash reporting — keeping the App working and understanding which features are used — and on contract (Article 6(1)(b)) for processing your purchase. Where India's Digital Personal Data Protection Act, 2023 applies, equivalent grounds apply.
Because we operate no accounts and hold no personal profile, we generally cannot identify you or link data to you. You can:
- Delete everything held on the device — clear the App's storage in Android Settings, or uninstall the App. This removes transcripts, models, settings and counters immediately. Step-by-step: How to Delete Your Data.
- Reset your analytics identity — reset the advertising identifier in Android Settings.
- Request information or deletion — write to [email protected]. Note that we may be unable to locate records relating to you, precisely because we do not collect identifying data.
Analytics cannot currently be turned off inside the App. If you would prefer it were, tell us at [email protected]; we would rather add the switch than have you uninstall.
5. Backups
If you have Android Auto Backup enabled, Android copies part of the App's data to your Google account. The App restricts this to a single preferences file containing usage counters. Your transcript history and your speech models are explicitly excluded from backup, so your transcripts are never copied to Google Drive.
6. Data retention
Data on your device stays until you delete it; the transcript history is capped at the 200 most recent entries and older ones are discarded automatically. Analytics data is retained by Google for up to 14 months. Crash reports are retained for up to 90 days.
7. Children
The App is not directed at children under 13, and we do not knowingly collect data from them. If you believe a child has provided us with data, contact us and we will delete it.
8. International transfers
Google processes analytics, crash and configuration data on servers that may be located outside your country, including in the United States, under Google's own safeguards. See https://policies.google.com/privacy.
9. Changes to this policy
If this policy changes materially we will update the effective date above and, where the change affects what we collect, note it in the App's release notes.
10. Trademarks
WhatsApp is a trademark of WhatsApp LLC / Meta Platforms, Inc. WA Transcriber is not affiliated with, endorsed by or connected to WhatsApp LLC or Meta Platforms, Inc. The App only reads audio files that you explicitly share with it.
11. Contact
Binaryscript Private Limited
Raipur, Chhattisgarh, India
[email protected]