Privacy Policy

Tolkimo · draft of 2026-06-10 · replace the contact and operator details before public release

What the app does

Tolkimo translates a live conversation: one participant's voice is streamed to our server and to an external translation service, and the translated speech is played on the other participant's device.

Data we process

• Microphone audio — only while the talk button is held (or while voice activation is on). Streamed to our server and to the Google Gemini API for translation.
• Text transcripts — listener-language text shown on the participants' screens. If the optional AI quality audit is enabled by the operator, short transcript snippets may also be sent to a configured OpenAI-compatible provider to detect translation misunderstandings. This audit is disabled by default.
• Technical data — room code, selected languages, server service logs.

What we do NOT do

• We do not store audio or transcripts after the conversation ends.
• We do not require accounts, names, phone numbers or e-mail.
• We do not use conversation content for advertising.
• We do not identify anyone by their voice.

Third-party services

Translation is performed by the Google Gemini API; audio processing is governed by Google's terms. Synthesized speech is watermarked with SynthID on Google's side.

If AI quality audit is enabled, text transcript snippets are processed by the configured OpenAI-compatible provider for live quality analysis; audio is not sent to that audit provider. By default, generated audit prompts are not stored in server logs; if audit logging is explicitly enabled, short derived summaries may be kept in operational logs for debugging.

Participants' responsibility

The person who starts a conversation must make sure the other participant understands that the conversation is translated by AI and their voice is being processed. In some jurisdictions recording or processing another person's speech without consent is prohibited.

Contact

Data questions: privacy@tolkimo.com