Official documentation
HoldSpeak SECURITY.md: Security and privacy (GitHub repository)
karolswdev, GitHub (karolswdev/HoldSpeak), 2026
Read the originalRead on October 25, 2026
- What it is
- The security and privacy document in the HoldSpeak repository, linked from its README. It lists where each kind of data lives on the user's machine and every point where data can leave it, including browser mic capture and the web and native dictation recovery copies.
- How much weight it carries
- The maker's own document, read from the default branch on 2026-10-25. It describes how the app handles data rather than a legal privacy policy, and it gives no number of days for meeting transcripts or for the local database. The statements are the maker's own and have not been audited.
Facts taken from this source
- HoldSpeak's security document says that for browser mic capture the WAV goes to the user's own hub, local Whisper transcribes it, and HoldSpeak does not keep the audio. The document gives no number of days.
“The WAV goes to your own hub, local Whisper transcribes it, and HoldSpeak does not keep the audio (16 MB cap).” (Section 4, Egress points, Browser mic capture row)
- HoldSpeak's security document says a pending browser voice capture is kept only while transcription has not confirmed its text, and that a successful transcription deletes it. The document gives no number of days.
“One bounded WAV per scope, kept only while transcription has not confirmed text. A successful transcription deletes it.” (Section 1, Data classes, Web pending voice capture row)