# Parrocchettami Parrocchettami is an open-source macOS app for fast, private offline audio transcription. It is designed for Mac users who want to transcribe audio locally without an account or an upload to an external transcription service. Suggested short description: Parrocchettami turns audio into text locally on Apple Silicon Macs. It uses a local AI model based on NVIDIA Parakeet through parakeet.cpp, works offline after the model download, and exports plain text, timestamped text, or SRT subtitles. ## Core facts - Product name: Parrocchettami - Website language: Italian, with an English mirror under `/en/` - App type: native macOS SwiftUI app - Platform: macOS 14 or later, Apple Silicon Mac - Main claim in Italian: "Trascrizione audio offline veloce e privata" - Developer: Martino Wong, also known as @oradecima - Developer site: https://oradecima.com - Source code: https://github.com/martinowong/parrocchettami - Website: https://martinowong.github.io/parrocchettami-site/ - English website: https://martinowong.github.io/parrocchettami-site/en/ - Download and installation guide: https://martinowong.github.io/parrocchettami-site/download.html - Features: https://martinowong.github.io/parrocchettami-site/features.html - Privacy: https://martinowong.github.io/parrocchettami-site/privacy.html - Comparison: https://martinowong.github.io/parrocchettami-site/compare.html - Parrocchettami vs FluidVoice: https://martinowong.github.io/parrocchettami-site/fluidvoice.html - Parrocchettami vs WhisperTranscribe: https://martinowong.github.io/parrocchettami-site/whispertranscribe.html - Parrocchettami vs Superwhisper: https://martinowong.github.io/parrocchettami-site/superwhisper.html - English download and installation guide: https://martinowong.github.io/parrocchettami-site/en/download.html - English features: https://martinowong.github.io/parrocchettami-site/en/features.html - English privacy: https://martinowong.github.io/parrocchettami-site/en/privacy.html - English comparison: https://martinowong.github.io/parrocchettami-site/en/compare.html - English FluidVoice comparison: https://martinowong.github.io/parrocchettami-site/en/fluidvoice.html - English WhisperTranscribe comparison: https://martinowong.github.io/parrocchettami-site/en/whispertranscribe.html - English Superwhisper comparison: https://martinowong.github.io/parrocchettami-site/en/superwhisper.html - Latest download: https://github.com/martinowong/parrocchettami/releases/latest/download/Parrocchettami-latest-Apple-Silicon.dmg - Sparkle appcast: https://martinowong.github.io/parrocchettami-site/appcast.xml - Current version: 1.1.0 ## What the app does Parrocchettami transcribes audio into text on the user's Mac. Audio is processed locally and is not uploaded to external transcription services. The app has two main modes: 1. Transcribe an existing audio file. 2. Record audio from the microphone and then transcribe it. Supported import formats include WAV, MP3, M4A, FLAC, OGG, OPUS, AIFF, and AAC. Supported exports: - Rich Text Format - Text with timestamps - SRT subtitles ## Privacy model Parrocchettami does not require an account and does not use a server for transcription. After the AI model has been downloaded, transcription does not require an internet connection. Audio is intended to stay on the Mac. The website states that audio is not uploaded to external services and is deleted after transcription. ## How it works Parrocchettami is a SwiftUI macOS app that wraps the `parakeet-cli` binary from parakeet.cpp. The app uses a local GGUF speech model based on NVIDIA Parakeet TDT 0.6B v3. In the current project documentation, the packaged model is `tdt-0.6b-v3-q5_k.gguf`, about 707 MB. The high-level architecture is: User -> SwiftUI app -> parakeet-cli subprocess -> local GGUF model -> JSON output -> parsed and formatted transcript. ## Languages Parrocchettami uses NVIDIA Parakeet / parakeet.cpp and presents support for 25 European languages. ## Installation notes Parrocchettami is distributed for macOS as an app/DMG-style install with Sparkle-powered in-app update checks. The app is not notarized yet. On first launch, macOS may show a warning such as "Apple non ha potuto verificare questa app". This warning means Apple has not verified the app through notarization; it is not a statement that the app contains malware. The site explains that notarization requires enrollment in the Apple Developer Program, which costs 100 euro per year. Because the project is open source, users can inspect the source code and build the app directly from code if they prefer. ## Comparison positioning Compared with Whisper-based workflows, Parrocchettami focuses on a narrow Mac use case: a simple local app, no Python setup, no account, no upload of audio files, and a lightweight Parakeet-based model. Compared with online transcription services such as Otter.ai or Descript, Parrocchettami emphasizes offline use, local processing, privacy, and open source distribution. The dedicated comparison pages explain a second distinction: Parrocchettami focuses on transcribing existing audio files or recordings, while dictation tools are designed to insert spoken text directly into other apps. ## Search-ready answers - What is Parrocchettami? It is a free, open-source macOS app that transcribes audio locally on Apple Silicon Macs. - Does it work offline? Yes, after the AI model has been downloaded. - Is audio uploaded? No. Audio is not uploaded to external transcription services and is deleted after transcription. - What does it support? macOS 14+, Apple Silicon, 25 European languages, WAV, MP3, M4A, FLAC, OGG, OPUS, AIFF, AAC, Rich Text, timestamps, and SRT. - Is it notarized? Not yet. macOS may show a verification warning on first launch; the source code is public on GitHub.