The short version: Your audio is written to temporary storage during processing and permanently deleted the moment analysis is complete. We never store, sell, or share your recordings.
What we collect
Email address — used solely for account authentication via Supabase.
Audio recordings — uploaded or recorded by you for analysis. Written to temporary server storage during processing and permanently deleted once analysis is complete. Never retained long-term.
Behavioural analysis results — transcripts, speaker patterns, dimension scores, and insights generated from your recordings. Stored in your account so you can track trends over time.
Voiceprint embedding (optional) — if you choose to use the voice training feature, a mathematical representation of your voice (not a recording itself) is stored to help identify which speaker is you in uploaded files. This is entirely optional, can be skipped, and can be deleted at any time.
What we do not collect
We do not retain raw audio after analysis is complete.
We do not sell, rent, or share your data with third parties for advertising or marketing.
We do not track your browsing activity or behaviour outside of the Mirror extension.
We do not record any conversation automatically — recording always requires an explicit action from you.
How your data is processed
When you submit a recording, it is sent securely to our processing server. The following steps occur:
Audio is written to temporary server storage and transcribed using the Groq Whisper API (Groq's privacy policy applies to transcription).
Speaker diarization and voice analysis are performed using pyannote.audio models running on our server.
Behavioural insights are generated using Groq's language model API.
The audio file is permanently deleted from the server immediately once processing is complete.
Analysis results (not the audio) are stored in your account database hosted on Supabase, a GDPR-compliant cloud database provider.
Google Meet integration
When using Mirror with Google Meet, the extension can capture audio from your meeting tab only when you explicitly click the Record button. Before your first recording, the extension displays a consent prompt reminding you that you are responsible for ensuring all meeting participants are aware the conversation is being recorded. Audio is never captured automatically or without your knowledge. The same data handling rules apply: audio is processed and permanently deleted; only the resulting insights are stored.
Third-party services
Supabase — authentication and database storage. Privacy policy.
Groq — speech transcription and language model inference. Privacy policy.
HuggingFace — model hosting and inference infrastructure. Privacy policy.
Your rights
You can delete all your data at any time from within the extension: open the Mirror panel → Account → Delete account. This permanently removes your analysis history, voiceprint, and account from our systems. You can also contact us directly to request data deletion.
Data security
All data is transmitted over HTTPS. Account authentication uses industry-standard JWT tokens managed by Supabase. Audio files are stored in temporary server memory only during processing and are never written to persistent disk storage.
Changes to this policy
If we make material changes to how we handle your data, we will update the effective date of this policy. Continued use of Mirror after changes are posted constitutes acceptance of the updated policy.