Ghost Pepper drops local hold-to-talk dictation

Ghost Pepper solves the latency and privacy issues of cloud dictation by bringing Whisper and Qwen directly to the desktop.

• –Powered by WhisperKit for high-speed, local transcription on Apple Silicon
• –Employs a local Qwen 3.5 LLM to automatically strip "ums," "uhs," and self-corrections
• –Simple hold-to-talk UX mimics walkie-talkie interactions, reducing UI friction during dictation
• –Fully open-source and privacy-first, requiring no cloud APIs or data transmission