April 22, 2026
Spice or Spam? HN can’t decide
Show HN: Ghost Pepper Meet local meeting transcription and diarization
Ghost Pepper: private voice notes go viral while HN shouts ‘repost’
TLDR: Ghost Pepper is a free Mac app that transcribes speech and meetings entirely on your computer, rocketing to #1 on Hacker News. Comments split between praise for privacy and speed, jokes about 'diarrheaization,' and pushback that it’s a too‑soon repost, making a spicy debate over spotlight vs spam.
Ghost Pepper, a free Mac app that turns speech and meetings into notes without using the cloud, just lit up Hacker News — and the comments are even hotter. Fans gush over the privacy-first, on‑your‑machine setup and the hold‑to‑talk speed, while the pepper logo gets its own fan club.
Then the drama kicks in. One side jokes about “diarrheaization,” roasting the jargon behind speaker labeling, and another side yells repost, pointing to a near‑identical Show HN from just 16 days ago with a link. “Seems like spam,” snipes one commenter, as others argue indie devs deserve a second shot when momentum hits.
Between quips, people highlight the practical wins: meeting recordings saved as local markdown, AI “cleanup” that deletes filler words, and 50+ language support — all on device, no account, no tracking. The app’s GitHub repo shows it’s open source and free, which only fuels the hype.
Verdict from the crowd? Equal parts wow, private and useful and hey, stop gaming the front page. If you like your voice notes extra spicy and offline, this is your snack; if you hate reruns, you’re reaching for milk. Either way, the pepper stole the show — logo included. Twice as hot.
Key Points
- •Ghost Pepper is a free, open-source macOS app for on-device voice dictation and meeting transcription, requiring Apple Silicon (M1+) and macOS 14+.
- •All features run locally with no cloud uploads, accounts, analytics, or telemetry; a privacy audit claims verification via AI code review.
- •Speech-to-text uses WhisperKit/FluidAudio locally; cleanup uses on-device Qwen LLMs via LLM.swift; meeting outputs are saved as local Markdown.
- •Available speech models include Whisper tiny.en, Whisper small.en (default), Parakeet v3, and Qwen3-ASR 0.6B (macOS 15+); cleanup models are Qwen 3.5 0.8B/2B/4B with size/speed trade-offs.
- •Installation is via a downloadable DMG on GitHub; the site notes a Gatekeeper “Open Anyway” step. The project highlights community reception with a “#1 on Hacker News” badge.