macparakeet
Local-first macOS app for system-wide dictation, meeting recording, and batch media transcription on Apple Silicon.
Transcribe Agile standups, record meetings, and dictate project notes entirely on your Apple Silicon Mac.
MacParakeet is a local-first macOS app that combines system-wide dictation, multi-speaker meeting recording, and media transcription. It processes speech directly on Apple Silicon without cloud dependencies, making it ideal for Agile teams and project managers who need private meeting notes. You can record calls, drop recording files, or trigger dictation hotkeys across any desktop application.
What it is
The app records system audio and microphone inputs simultaneously to generate live transcripts of project meetings and sprint reviews. It offers local speech-to-text models like Parakeet, WhisperKit, and Nemotron, alongside audio export options in formats such as Markdown, DOCX, and SRT. You can also connect optional AI providers for automated transcript summarization and selected-text rewrites.
Who it's for
It suits project managers, Scrum masters, and product owners who want accurate, offline meeting logs without exposing sensitive team discussions to cloud vendors. It is not suitable for users on Windows, Linux, or Intel-based Macs. Teams looking for automatic cloud syncing without running local models will also find it restrictive.
Key features
System-Wide Voice Dictation
Press a global hotkey or chord to speak and insert formatted text directly into any macOS application.
Dual-Audio Meeting Capture
Capture microphone and system audio together with live transcript previews and manual note-taking capabilities.
Local Calendar Integration
Read local EventKit calendar events from Exchange or M365 to trigger opt-in meeting recording countdowns.
Batch File and URL Processing
Drag local media files or paste video and podcast links to generate detailed transcripts with speaker labels.
CLI Agent Automation
Use macparakeet-cli to run automated podcast searches, media downloads, and batch transcriptions from your terminal.
Up and running in 3 steps
Our honest verdict
What's great
- Runs entirely on-device using Apple Silicon Neural Engine for full privacy.
- Combines dictation, dual-channel meeting capture, and batch file transcription in one interface.
- Includes calendar integration for event-driven meeting capture and local reminders.
- Supports automated CLI commands and exports to 8 standard formats including Markdown and JSON.
Where it falls short
- Requires an Apple Silicon Mac (M1 or newer) and cannot run on Windows or Intel Macs.
- Alternative speech models like WhisperKit and Nemotron require multi-gigabyte local downloads.
- Third-party AI summarization features require your own external API keys or local LLM setup.
- Cohere speech engine runs in batch mode only without live previews or speaker timestamps.
Pricing & license
MacParakeet is free and open-source software with no subscription or license fees. The only costs are your own Mac hardware and optional API usage fees if you choose to connect third-party LLM providers.
Open remote AI-training roles
879 open remote AI-training roles, updated November 15, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
These reviewed AI tools do the same job instantly — from our scored directory.
Use it, or have us set it up
FAQ
Does MacParakeet require an internet connection to transcribe speech?
No, core speech recognition runs completely on your Mac. Internet access is only needed if you explicitly enable opt-in cloud AI features or paste external media links.
Can it automatically record Zoom or Teams meetings?
It reads local macOS Calendar events to offer countdown reminders or opt-in auto-start recording, but actual meeting recording runs locally via macOS audio capture.
Which languages does the default speech engine support?
The default Parakeet v3 engine auto-detects among 25 European languages. For languages like Japanese, Korean, or Chinese, the app defaults to WhisperKit on first setup.
