Speaker Scribe

Local transcription for Apple Silicon

A private transcript, with every speaker in view.

Speaker Scribe turns recordings into readable, speaker-separated transcripts. The speech stack runs on your Mac—no account, API key, or audio upload.

Download for macOSApple Silicon · macOS 13+Read the source
  • Signed & notarized
  • 1.05 GB
  • MIT licensed
The actual Speaker Scribe interface showing a completed three-speaker transcript, speaker names, model selection, and export controls
Actual interfaceA real three-speaker transcript, ready to inspect and export.

THE LOCAL PATH

Your recording never needs a round trip.

Decode, speech recognition, voice grouping, review, and export all happen on-device. Optional model files download only when you ask.

Read the exact privacy boundary →
Speaker Scribe local processing flowA recording moves through models on the Mac into a local transcript and export. Nothing in the flow uploads the recording.YOUR AUDIOON-DEVICE MODELSLOCAL TRANSCRIPT

DOWNLOAD → OPEN → CHOOSE

Download once. Choose your model when it opens.

The signed app already includes Small, a practical offline default. Keep it, or choose another model from the real picker inside Speaker Scribe.

  1. 01

    Downloadthe reviewed Apple Silicon app.

  2. 02

    Openit with Small already available.

  3. 03

    Choosefor speed or accuracy before transcribing.

BUILT TO BE CHECKED

It separates voices. It does not identify people.

Speaker Scribe can detect that distinct voices appear in a recording. You decide that one is Dana. It can also mishear words or split a speaker, so playback and verbatim text stay close at hand.

Speaker 1 → DanaSpeaker 2 → PriyaSpeaker 3 → Marcus

READY WHEN THE RECORDING IS

Your transcript belongs on your Mac.

Download the signed build. Small is ready; every other model is one in-app choice away.

Download for macOSApple Silicon · macOS 13+Inspect it on GitHub ↗