Does transcription run locally?
Yes. SaidVault uses a Whisper model that runs on your device. Audio is processed locally; the recording, the transcript, and the metadata never leave your machine during transcription. Once a model is downloaded, you can work fully offline.
What file formats are supported?
MP3, MP4, M4A, WAV, MOV, WEBM, OGG, and AVI for both audio and video. You can also paste a direct media URL and SaidVault will fetch and transcribe it.
Can I export to PDF or TXT?
Yes — TXT, PDF, and SRT. PDF exports include a header with reference date, file size, source SHA-256 hash, model used, language, duration, and word count.
Can I assign speakers?
Yes. You create named speakers and assign them to transcript segments manually. Labels carry through to exports. Automatic speaker detection (diarization) isn't part of SaidVault today.
Does it support dictation?
Yes. Hold a global shortcut (default ⌘⇧V, configurable) anywhere on macOS to start a push-to-talk capture. The transcript lands on your clipboard ready to paste. A floating HUD shows Listening, Transcribing, and Copied states.
What languages does the interface support?
English, Polish, German, Dutch, French, and Spanish. The interface labels, toasts, and export metadata are all translated. Whisper itself supports many more languages for the audio side.
Is Windows coming?
macOS is the active platform. A Windows port is planned after the macOS v1 release. We're keeping cross-platform-sensitive parts of the code (recorder, shortcuts) explicit so the port stays clean.
Where is my data stored?
Transcripts and voice notes live in a local SaidVault library on your Mac. You control where exports save. There is no SaidVault cloud account.