Ren STT v1.0.0

Local speech-to-text for macOS. Superwhisper replacement — runs entirely on your hardware.

Install

One-liner:

curl -fsSL https://raw.githubusercontent.com/phyter1/ren-stt/main/get.sh | bash

Or download the DMG: grab RenSTT.dmg below, drag to Applications, open. First launch walks you through setup.

What's in v1.0.0

• Parakeet TDT on MLX — 0.6B (fast) or 1.1B (accurate) model selection
• Option+Space to record, Escape to cancel — hotkey intercepted at system level, no extra spaces
• Automatic punctuation — commas, periods, question marks, capitalization via ONNX punct_cap_seg model (~65ms)
• Filler word removal — strips "uh", "um", "hmm", "er" from transcriptions
• Menu bar icon — waveform (SF Symbols), red when recording, orange when transcribing
• Three install modes — Standalone (server + client), Server only, Client only
• Network architecture — one Apple Silicon server, clients on any Mac
• Server watchdog — auto-restarts if server process dies
• Single-launch permissions — Accessibility auto-detected, no manual relaunch needed
• Login Item — starts on boot

Performance (M1 Pro, 1.1B model + punctuation)

Audio	Latency
2s	~250ms
6s	~300ms
16s	~530ms
47s	~1.9s

Requirements

• macOS 12+, Python 3.9+
• Apple Silicon for server/standalone
• sox (brew install sox) for client