The Privacy-First Way to Transcribe Signal Voice Notes
Signal = Privacy
If you use Signal, you probably care about End-to-End Encryption (E2EE). You use it because you don't want Meta (WhatsApp) or Apple reading your messages.
But when you get a voice note, you have a problem. To transcribe it, most people use online tools or bots. Using a bot breaks the chain of custody. If you forward a secure Signal voice note to a Telegram bot for transcription, you have just compromised your security.
The "Air-Gapped" Approach
The safest way to transcribe audio is on-device or via a stateless API that you control.
How TalkFlow Fits In
TalkFlow is an iOS Shortcut, not a cloud app.
- You receive the file in Signal.
- You share it to the Shortcut.
- The Shortcut sends audio to OpenAI (via API) and gets text back.
- No history is saved.
The Workflow
- Open the voice note in Signal.
- Tap the Share icon.
- Select TalkFlow.
- Get the text.
- (Optional) Paste the text back into Signal as a "Note to Self" or reply.
Why not on-device dictation?
Apple's on-device dictation is good, but it can't process files. It only listens to the microphone. To transcribe a file locally, you need a heavy model. TalkFlow bridges this gap using the API for speed and accuracy while maintaining a "zero-retention" policy.
Keep your conversations private, even when they are audio.