Privacy Policy - Clarion: Speaking Practice

Last updated: October 9, 2026

1. Introduction

EasyTech Agency ("we," "our," or "us") operates the Clarion mobile application for iOS and Android, formerly called Vocal Coach (the "Service" or "App"). This page informs you of our policies regarding the collection, use, and disclosure of personal data when you use our Service.

In short: you never create an account with your email or name. Your recorded takes, free or Pro, are sent to our server to be transcribed word for word (so filler words like "um" can be counted) and analyzed, then not kept. Without a connection, takes are analyzed on your device instead. We do not sell your data.

This policy replaces the previous Vocal Coach policy. Unlike earlier versions of the App, Clarion sends your recorded takes to our server, as described in section 2.4.

2. Data We Collect and How We Use It

2.1 Anonymous Account

On first launch, the App signs you in silently with Firebase Anonymous Authentication (Google). This creates a random identifier (UID) for your installation. We never ask for your email, name, phone number or a password, and the UID is not linked to your Apple ID or Google account. It is used to authenticate requests to our server, apply daily usage limits, link your subscription (RevenueCat) and group analytics events (PostHog). The App also creates a random install identifier, stored in your device's Keychain (iOS) or in the App's private storage (Android, where it is the only item included in Google backups), used only to make sure the one-time onboarding voice check (section 2.4) runs once per device and to apply daily limits.

2.2 Microphone and Speech Recognition

Clarion records your voice while you practice, with your permission (the microphone and, on iOS, speech recognition prompts). Your recording is also turned into words with timings by Apple's Speech framework, for live meters while you speak and as a fallback when our server cannot be reached (section 2.4). On supported iPhones this happens on your device; when on-device recognition is not available, Apple may process the audio on its servers under Apple's Privacy Policy. Pitch and volume are measured on your device. Without a connection, pace, pauses and filler words are measured on your device too, from Apple's words (filler words are then estimated, and the App says so).

On Android, live words and the offline fallback use your phone's on-device speech recognition service only. Clarion asks Android for on-device recognition and never uses the server-based recognizer, so the audio reaches our server (section 2.4) and nowhere else.

2.3 Text Sent to Our Server (no audio)

In the following cases, the App sends text only, never audio, to our server:

• Free path levels: after your take has been transcribed (section 2.4), its transcript (the words and their timings), the practice prompt and a few measurements, so that our server can judge things like whether you made a clear point or gave a specific example.
• Mock interview answers (Pro): the on-device transcript of each answer, with the role and optional company you entered and the questions so far, so the interviewer can ask the next question or follow up on what you said.
• Long Talk run-throughs (Pro, over 2 minutes): the transcript of your rehearsal (up to its first 1,500 words).

2.4 Audio Sent to Our Server

Your recording is uploaded to our server in these cases only:

• The onboarding voice check (once per device): a short take of about 20 seconds (25 seconds at most), so you can see your own real feedback before you start, with filler words counted from a word-for-word transcript.
• Free path level (your first level, one take a day, up to 2 minutes a take): each take is transcribed word for word so filler words can be counted; the analysis is then completed with the text described in section 2.3.
• Pro reps (path levels and Daily reps, up to 2 minutes), for a word-for-word transcript and written coaching.
• Pro Talk run-throughs of 2 minutes or less.
• Mock interview (Pro): the recording of your best answer only (up to 2 minutes), for word-for-word coaching in the debrief.

Our server runs on Google Cloud Run in the European Union (region europe-west1, Belgium). There, your audio is transcribed by ElevenLabs (speech-to-text). The transcript is judged by TypeSafe's Jev text-analysis service, and short coaching sentences (or, for the voice check, an improved version of your answer) are written by Google Gemini through Vertex AI. The results are sent back to your phone. Takes are uploaded only if they contain some speech. If there is no connection, or our server is unavailable, the take is not sent and is analyzed on your device with Apple's speech recognition (section 2.2).

We do not store your audio or your transcripts on our servers. They are processed in memory to produce your feedback and discarded once the response is sent. We never use them to train AI models.

2.5 "Hear the Future You" (Voice Cloning, With Your Consent)

During onboarding, Clarion offers to play an improved version of your voice-check answer in your own voice. This only happens if you explicitly agree on that screen. If you do, your voice-check recording is used to create a temporary voice with ElevenLabs (instant voice cloning), which reads the improved answer once. The cloned voice is deleted immediately after that single reading. It is used for nothing else, never shared, never reused and never used by us to train any model. This happens at most once per device. If you do not agree, no cloned voice is created (the App may use a standard coach voice instead, or no audio at all).

2.6 Mock Interview Voice

In a Mock interview, the interviewer (Maya) speaks with a synthetic voice generated by ElevenLabs (text-to-speech) from the text of her lines. Her lines are written by Google Gemini, based on the role, company and answer text described in section 2.3. To avoid generating the same audio twice, the audio of Maya's lines is cached in our Google Cloud Storage, under a name derived from the text itself. This cache contains no recording of your voice and no identifier of you; however, a follow-up question may briefly quote words from your answer.

2.7 What Our Server Keeps

Our server stores no audio, transcripts or feedback. It keeps only:

• Daily usage counters keyed by your anonymous UID and install identifier (Google Cloud Firestore), used to enforce daily limits (for example, the number of free takes transcribed each day). They contain counts only, no content. They expire automatically after about two days.
• A record that your UID and install identifier have used the one-time onboarding voice check (a date only, no content).
• Technical logs (for example, which feature was called, its estimated cost, your anonymous UID and error messages), kept for a limited time for security and troubleshooting.
• The interviewer audio cache described in section 2.6.

2.8 Data Stored on Your Device

• Your recordings (the most recent reps, plus a few kept for "your voice then vs now"). They are excluded from iCloud, Google and device backups, and older audio is deleted automatically.
• Your progress: ranks, levels, stars, XP, skills, streaks and the measurements of each rep
• Talks you write or paste for Talk run-throughs
• Your onboarding choices, mock interview setup (role, company, seniority), coach voice and reminder settings

You can delete all of it at any time in Settings > Delete my data: this wipes your progress, recordings, talks, interviews and profile from the device, signs out the anonymous account (a new anonymous identifier is created) and resets the analytics identifier. Uninstalling the App also removes this data.

2.9 Notifications

Daily reminders are scheduled locally on your device, only if you allow notifications. We do not run a push-notification server.

3. Subscriptions and Payments

In-app purchases and subscriptions are processed by Apple (iOS) or Google Play (Android) and managed through RevenueCat, a third-party service. RevenueCat may collect:

• An anonymous app user identifier (your anonymous UID) to manage your subscription
• Your in-app purchase history and subscription status
• No payment information (handled directly by Apple or Google Play)

Our server asks RevenueCat whether your UID has an active Pro subscription before running Pro features. For more information, see the RevenueCat Privacy Policy.

4. Analytics

We use PostHog to understand how the App is used and improve it. We send product events such as onboarding steps, paywall views and purchases, and rep, interview and run-through completions with their measurements (for example duration, pace, filler count, stars). Events are linked to your anonymous UID, not to your name or email. No audio and no transcript text are ever sent to analytics. For more information, see the PostHog Privacy Policy.

5. Advertising Attribution (App Tracking Transparency, Android advertising ID)

Clarion uses the TikTok Business SDK and the Meta (Facebook) SDK to measure which advertising campaigns help people discover the App. The events sent are limited to: onboarding completed, free trial started and subscription started (with the product and price).

• The App may ask your permission to track through Apple's App Tracking Transparency prompt. Only if you allow it is your device's advertising identifier (IDFA) shared with TikTok and Meta.
• If you decline (or are never asked), no advertising identifier is shared and we do not track you across other companies' apps and websites. The events above may still be sent without that identifier, for aggregated measurement such as Apple's SKAdNetwork.
• You can change your choice at any time in iOS Settings > Privacy & Security > Tracking.

On Android, the App first shows its own screen explaining this measurement. Only if you tap Allow do the TikTok and Meta SDKs start; they may then read your Android advertising ID and send the events above, so we can measure which of our campaigns bring installs and subscriptions. If you tap Not now, nothing is sent. The App also reads the Google Play install referrer (the campaign tag of the Play Store link used to install the App) for the same purpose. You can reset or delete your advertising ID at any time in Android Settings > Privacy > Ads. We do not show advertising in the App.

See the TikTok Privacy Policy and the Meta Privacy Policy.

6. Service Providers

We rely on the following providers, each limited to the role described:

• Google (Firebase Authentication, Cloud Run, Firestore, Cloud Storage, Vertex AI / Gemini): anonymous sign-in, hosting of our server, usage counters, interviewer audio cache and coaching text (Firebase privacy, Google Cloud privacy notice)
• ElevenLabs: speech-to-text for uploaded audio, the interviewer's voice, and the temporary voice clone (with consent) (privacy policy)
• TypeSafe (Jev): text-only judgment of your transcripts
• Apple: speech recognition, purchases, App Tracking Transparency (iOS)
• Google Play: purchases and the install referrer (Android)
• RevenueCat: subscription management
• PostHog: product analytics
• TikTok and Meta: advertising attribution (section 5)

Our server is located in the European Union. Some providers (for example ElevenLabs, RevenueCat, PostHog, TikTok and Meta) may process data outside the European Union, and when Google's EU capacity is saturated, Gemini requests may be handled in other regions. Such transfers rely on appropriate safeguards such as the European Commission's Standard Contractual Clauses or the EU-U.S. Data Privacy Framework.

7. Data Sharing

We do not sell, trade, or rent your personal information. We share it only:

• With the service providers listed above, for the purposes described
• With your explicit consent
• To comply with a legal obligation
• To protect our rights or your safety

8. Data Retention

• Audio and transcripts sent to our server: not kept after the response is sent
• Cloned voice (if you consented): deleted right after one use
• Daily usage counters: expire after about two days
• On-device data: until you delete it (Settings > Delete my data) or uninstall the App
• Subscription, analytics and attribution data: as long as needed for these purposes, according to each provider's policy

9. Data Security

All traffic between the App and our server uses TLS encryption, and every request to our server must carry a valid Firebase token. Data on your device is protected by your device's built-in security features. Subscription data is secured by Apple and RevenueCat.

10. Your Rights

Under the GDPR and similar laws, you have the right to:

• Access the personal data we hold about you
• Delete your data: in the App with Settings > Delete my data, or by asking us
• Correct or receive a copy of your data (portability)
• Object to or restrict processing
• Withdraw your consent at any time (microphone, speech recognition, notifications and tracking can be changed in your device Settings)
• Lodge a complaint with a supervisory authority (CNIL in France)

To exercise these rights, contact us at contact@easytech-agency.net. Because we never know your name or email, we may need your help to find the data linked to your anonymous account; the quickest way to erase what the App holds is Settings > Delete my data.

11. Children's Privacy

The App is not directed to children under 13 (or under 16 in countries of the European Union where that is the age of digital consent). We do not knowingly collect personal information from children. If you believe a child has used the App and data was collected, please contact us and we will delete it.

12. Changes to This Policy

We reserve the right to modify this Privacy Policy at any time. Changes will take effect upon publication. We encourage you to review this page periodically.

13. Contact

If you have any questions about this Privacy Policy, contact EasyTech Agency at:
contact@easytech-agency.net