SpeakSheet Privacy Policy

Effective: May 20, 2026 · Updated: [TO CONFIRM: date the beta signup form goes live] to cover beta signups

1. Who we are

SpeakSheet is operated by Dennis Dyallo, Sweden. Contact: [email protected]. We are the data controller for the personal data described in this policy.

2. Quick summary

3. Data we process

Category What Why Where it goes Retention
Microphone audio Short voice clips (≤5 min) Transcribe speech → chord sheet Sent to OpenRouter → Google Gemini 2.5 Flash for transcription Not stored by SpeakSheet; OpenRouter/Google retention per their terms
Chord/lyric text The text content of songs you create or edit Generate chord sheet (dictation) and apply AI edits Sent to OpenRouter → Anthropic Claude Not stored by SpeakSheet; OpenRouter/Anthropic retention per their terms
Shared songs The chord sheet payload of any song you tap "Share" on Generate a short-link recipients can open Stored in SpeakSheet backend at speaksheet.app Indefinite — until you contact us to delete
Beta signup The name, email address and platform choice (Android / iOS) you enter in the beta signup form on speaksheet.app Contact you about Android/iOS beta access and administer the beta. Lawful basis: consent (GDPR Art. 6(1)(a)), given by submitting the form; withdrawable at any time Stored in the SpeakSheet backend at speaksheet.app. Not shared with any third party [TO CONFIRM: retention period for beta signup records — must match the same figure in the Beta Terms], or until you ask us to delete it, whichever is sooner
Saved songs + setlists Your local song library Local app function only Your device only (localStorage) Until you uninstall the app
Crash reports Stack traces, page URL, browser/OS, app version, device model (NO chord/lyric content) Diagnose crashes Sentry 90 days (Sentry default)
Performance & Web Vitals URL of the page/route, page-load timing, Web Vitals (LCP, FCP, CLS, INP), and resource load timings — sampled at 10% of sessions Diagnose slow pages and regressions Sentry 90 days (Sentry default)
Website usage analytics Page views, feature taps, anonymous session counts, and truncated JS error messages/stack snippets (no chord/lyric content) (IP processed only to derive country and a daily anonymous session hash, then discarded; no cookies, no persistent user identifier) Product improvement for website users Umami at analytics.dyallo.se — website only; not loaded by native Android or iOS builds Indefinite (aggregate only)

We do NOT process:

4. Third parties

Service Purpose Country Their privacy policy
OpenRouter AI request routing (Anthropic + Google) USA https://openrouter.ai/privacy
Anthropic Chord/lyric generation + AI edit (via OpenRouter) USA https://www.anthropic.com/legal/privacy
Google (Gemini) Audio transcription (via OpenRouter) USA https://policies.google.com/privacy
Sentry Crash reporting USA / EU https://sentry.io/privacy/
Umami Anonymous website analytics only (self-hosted; not used by native Android or iOS builds) Sweden (self-hosted; no data sharing)

International transfers: Some processors are in the USA. We rely on Standard Contractual Clauses (SCCs) under GDPR Art. 46 for these transfers. By using SpeakSheet you consent to this transfer.

5. Your rights

Depending on where you live, you have rights including:

To exercise any right, email [email protected] — with the share code(s) you want deleted, and/or the email address you used on the beta signup form.

6. Children

SpeakSheet is intended for users 13 years and older. We do not knowingly collect data from children under 13. If you believe a child under 13 has used SpeakSheet, contact us and we will delete any associated data.

App Store listing: "Made for Ages 4+" rating is appropriate (no objectionable content), but the privacy policy explicitly limits the audience to 13+ to comply with COPPA / GDPR-K.

7. Security

8. Microphone permission

SpeakSheet uses the microphone only when you tap the dictate button. Audio is captured for the duration of your dictation, sent for AI transcription, and immediately discarded. We do not store or share the audio. Microphone access can be revoked at any time in iOS / Android system settings.

9. AI processing transparency

When you use dictation, your audio is transcribed by Google's Gemini AI model (via OpenRouter). When you use AI edit or generation, the chord/lyric text is processed by Anthropic's Claude model (via OpenRouter). These models do not retain your data per their terms of service. We do not use your data to train any AI model.

10. Changes to this policy

We may update this policy. Material changes will be announced in-app on the next launch. Continued use after notice constitutes acceptance.

11. Contact

Apple App Privacy / Google Play Data Safety mapping

For App Store Connect "Privacy" section + Play Console "Data Safety" form:

Field Answer
Audio Data — Microphone Yes — App Functionality — Not Linked to User
User Content — Other (chord/lyric text) Yes — App Functionality — Not Linked to User
Diagnostics — Crash Data Yes — App Functionality — Not Linked to User
Diagnostics — Product Interaction Yes — Analytics — Not Linked to User
Identifiers (User ID, Device ID) No (v1 has no accounts)
Contact Info — Name, Email Address No for the app itself — the beta signup form exists only on the website, and neither the Android nor the iOS build contains it or collects contact information. Name and email are processed on the website; see the "Beta signup" row in section 3
Health & Fitness No
Financial Info No
Location No
Sensitive Info No
Contacts No
Browsing History No
Search History No
Photos / Videos No (v1)
Tracking No