No account
SaidInk does not require a login or shared workspace.
Private transcription
Private transcription without sending audio to SaidInk.
SaidInk is designed for sensitive recordings: local files, local transcription after model download, no SaidInk account, no analytics SDKs, and user-controlled exports.
No analytics SDKs
The MVP build avoids advertising, analytics, and tracking SDKs.
Local models
Transcription runs locally after the selected model is downloaded.
Privacy boundary
Local-first does not mean pretending the internet never exists.
SaidInk's product promise is specific: it does not upload your audio or transcripts to a SaidInk server for transcription. Recordings, transcript edits, bookmarks, and local settings are stored on the device.
Network access can still happen for user-requested model downloads, model manifest/checksum validation, Apple StoreKit purchases and restores, explicit Apple system services, and the public website or support email you choose to use.
What stays local
The transcript workflow is built around your device.
Audio and transcripts
Recordings, imported audio, extracted audio, transcript rows, edits, and bookmarks live as local app data.
Model files
Fast, Balanced, and Accurate model files are downloaded only after the user starts the download.
Exports
You choose where exported TXT, timestamped TXT, bookmarked TXT, or SRT files go through iOS sharing.
Deletion
Settings includes local storage controls for deleting recordings, transcripts, model files, and app-local state.
Good for
Sensitive interviews, research, notes, and drafts.
Privacy matters most when the recording includes sources, students, clients, unreleased material, personal notes, or anything that should not be copied into a generic cloud transcript workspace.
Precise claim
Not a vague "no data" slogan.
Purchases, support email, model downloads, and website hosting have their own metadata boundaries. The important line is that SaidInk does not collect your audio or transcripts for cloud transcription.