Trust Center

Security, privacy and compliance you can read

Sprecho is a German GDPR-first voice-dictation product. The high-level summary lives here in the open; the DPA, sub-processor list, ISMS policies and audit evidence are one corporate-email request away.

  • Data classificationRestricted
  • ImpactModerate
  • RTO4h
  • RPO1h
  • RegionEU (Germany)

At a glance

The facts most procurement teams check first.

Data residency

EU only (Germany + Netherlands)

Encryption

TLS 1.2+ in transit · AES-256 at rest

Compliance

GDPR · hosting ISO 27001 (STRATO AG)

Default retention

14 days · zero-retention option for enterprise

No training on customer voice

Synthetic + open-source corpus only — your audio is never reused

Audit log

Every auth, admin and billing event recorded · JSONL export on Enterprise

MFA

Available for all users · required for admins on Enterprise

Penetration testing

First independent test scheduled Q3 2026

Browse the documentation

Request access to read the full content

Sprecho's Trust Center is open for procurement teams, security reviewers and customers. Submit your corporate email and we'll grant access — usually within minutes for verified domains.

Frequently asked

The five questions every procurement reviewer asks first. The full DPA and ISMS policies are one corporate-email request away.

  • Is Sprecho's environment multi-tenant?+

    Yes. Logical isolation is enforced at the database (tenant scoping via team_id / user_id), API (authentication plus authorization checks on every FastAPI route; admins cannot read other members' transcripts), storage (files keyed by tenant + user; authenticated download endpoint), and inference (tenant-aware queues; no cross-tenant batching) layers. Dedicated single-tenant deployments are available on request for Enterprise customers.

  • What telemetry does the Sprecho client app collect?+

    Operational telemetry only — no dictation content. Collected: application performance metrics (latency, error rates, vLLM fallback usage), feature usage events (which features activated, how often — never dictation content), crash reports (via Sentry, sanitised to strip sensitive context), auth events (login, MFA, password reset) for security monitoring. Never collected: dictation text or audio, browsing history, file contents, screen captures, contact lists, IDFA / advertising identifiers.

  • Which platforms does Sprecho support?+

    macOS (primary desktop, Tauri-built), Windows (Tauri-built, signed with Azure Trusted Signing), iOS (React Native, in TestFlight; App Store submission pending), Android (React Native), web admin portal at admin.sprecho.ai for organization configuration, enterprise dashboard at enterprise.sprecho.ai for multi-seat licence management.

  • How many languages does Sprecho support?+

    100+, with German as the highest-quality target. The Whisper model supports 100+ languages for transcription. Quality varies by language; German has the highest accuracy because the Sprecho SFT model is specifically tuned for German formatting (punctuation, paragraphing, capitalization, command-mode triggers). English second-strongest. Portuguese (PT and PT-BR) supported. Quality on minority languages can be evaluated during trial.

  • What macOS permissions does the desktop app require?+

    Microphone (mandatory, capture audio for dictation), Accessibility (used only for context inspection — reading the focused element or surrounding text to improve AI formatting accuracy; not used for text insertion, which uses clipboard plus simulated paste), Screen Recording (optional, only if the window-aware context feature is enabled), Automation (optional, specific apps only, for the per-app-tuning feature). All permissions are explained at first launch with clear opt-out paths; they can be revoked anytime in macOS System Settings → Privacy & Security.