Changelog
New features, improvements, and fixes shipped to MacParakeet.
July 2026
Custom Vocabulary Hotfix · v0.7.1
Paused recognition-time vocabulary boosting by default after short custom vocabulary lists could dominate final dictation output. Clean post-transcription vocabulary corrections still run.
Cohere Transcribe
Added Cohere Transcribe as an optional local batch speech engine for plain-text file transcription, with safer memory checks and a GPU/ANE compute setting.
Cleaner Meeting Transcripts
Final meeting transcription can use a cleaned microphone track to reduce speaker bleed while keeping the original recording available as fallback.
More Recoverable Meetings
Meeting finalization status is clearer, failed processing can be retried, and source audio artifacts now identify mic, system, mixed playback, and cleaned microphone audio more explicitly.
Voice Return Phrases
Voice Return now supports multiple trigger phrases, including multilingual equivalents, while still only triggering Return from phrases spoken at the end of dictation.
Library and Export Polish
Added bulk transcript export, local transcription title renaming, in-transcript find, improved reading controls, and cleaner PDF/DOCX exports.
June 2026
Parakeet Unified and Nemotron Beta
Added more local speech model choices, including Parakeet Unified for English and Nemotron beta live preview, while keeping Parakeet v3 as the practical default for broad app workflows.
Live Dictation Preview
Dictation can show a live preview while you speak on supported engines, with final pasted text still coming from the completed transcription.
Paste Almost Any Media Link
URL transcription expanded beyond YouTube to links supported by yt-dlp, including X, Vimeo, TikTok, Instagram, Facebook, Twitch, SoundCloud, Apple Podcasts, and more.
Podcast Transcription
Apple Podcasts links resolve directly to episode audio, and the CLI can search podcasts by name for walk-away transcription workflows.
Meeting Source Modes and Retention
Meeting recording can use mic plus system audio, mic only, or system only. Source audio can be kept, auto-deleted after a retention window, or removed after transcription.
Back-to-Back Meeting Recording
Start the next meeting without waiting for the previous one to finish transcription.
Configurable Dictation Undo
Keep the 5-second cancel countdown, shorten it to 1 second, or turn it off.
May 2026
YouTube Transcription Fix
Sorry about the YouTube transcription failure some fresh installs hit in recent builds. YouTube downloads now work again and affected helper copies are repaired automatically.
Audio Input Test Reliability
Fixed a 0.6.0 regression where some system-default USB or camera microphones could fail the Audio Input test.
Live Transcript Readability
Fixed speaker text colors that could be hard to read on light backgrounds.
Meeting Recording
Capture system audio and your microphone at the same time, then transcribe locally — your audio never leaves your Mac. Granola-style workflow, MacParakeet privacy.
Live Notes, Transcript, and Ask
Three tabs while you record: type notes alongside the audio, watch the transcript stream in, and ask questions about the meeting as it unfolds.
WhisperKit Multilingual Transcription
Optional local WhisperKit engine adds 73 more languages — Korean, Japanese, Chinese, Hindi, Arabic, Vietnamese, Thai, and more — for everything Parakeet doesn't cover. Still on-device, one-time download.
Guided Meeting Permissions
If Screen Recording or Microphone access is missing, MacParakeet now points you to the right setup path before recording.
Cleaner Transcript Defaults
Speaker diarization is now off by default for simpler transcripts. You can turn speaker labels on whenever you want them.
Meeting Reliability Polish
Onboarding, file transcription, AI provider setup, meeting recovery, and update handling all received reliability and usability improvements.
April 2026
Open-Source Release · GPL-3.0
MacParakeet went fully open-source. Full Swift source on GitHub under GPL-3.0. No paid tier, no community edition with features stripped out.
Prompt Library + Multi-Summary
Bring your own prompts. Run multiple summaries on the same transcript from different angles — meeting recap, action items, technical notes — without overwriting each other.
Library Redesign
Thumbnail grid, split-pane detail view, favorites, and a built-in video player for transcribed video files. Filter by source: all, YouTube, local, or favorites.
Multi-Conversation Transcript Chat
Keep multiple distinct chats per transcript instead of one rolling thread. Branch a new conversation when you change topics.
AI Formatter + LM Studio
Custom formatter prompts clean up dictation output your way. LM Studio joins Ollama as a local LLM option for users who don't want to touch a cloud API.
OpenAI-Compatible Providers
Connect any OpenAI-API endpoint — local servers, OpenRouter, custom proxies. Loopback URLs are treated as local for privacy.
Transcription Hotkeys + Auto-Copy
Bind a hotkey to start file transcription from anywhere. Finished transcripts auto-copy to your clipboard so the next paste is the result.
Idle CPU & GPU Drop
Killed perpetual SwiftUI animations that pegged the GPU when MacParakeet sat in the background. Idle resource use is now near-zero.
March 2026
Simplified Transcribe Screen
Two clear options: YouTube or local file. Cleaner layout, no clutter — just what you need to get started.
Better Transcription Progress
The pipeline view now shows directly without extra clicks. Smarter auto-scroll pauses when you scroll manually.
Copyable Error Messages
Copy the full error with one click if you need to report it.
Speaker Diarization
Know who said what — detect and label individual speakers in file transcriptions. Rename speakers inline, see per-speaker stats. Labels appear in all export formats.
AI Summary & Chat
Connect your own API key for one-click transcript summaries and conversational Q&A. Works with OpenAI, Anthropic, Ollama, OpenRouter, and any OpenAI-compatible endpoint. Entirely optional — the app works fully offline without it.
Chord Hotkey Support
Set modifier+key combos as your dictation trigger — Cmd+9, Option+Space, Ctrl+Shift+D, or any chord you like.
Private Dictation Mode
New toggle to disable dictation history saving entirely. Text is pasted as usual but nothing is stored.
Auto-Updates
MacParakeet now updates itself via Sparkle. No more manual DMG re-downloads.
Non-blocking Transcription
File transcription runs in the background with a progress bar — keep using the app while it works.
Discover Tab
Curated content about suppressed science, buried patents, and ancient wisdom. Rotating sidebar card, copy any card, and share your thoughts privately.
February 2026
Custom Hotkey Support
Set any key as your dictation trigger — not just Fn. Configure it in Settings.
Menu Bar Drag & Drop
Drop audio and video files directly onto the menu bar icon to start transcribing.
Voice Stats Dashboard
See your dictation stats — total words, sessions, and streaks.
Hide Pill Toggle
Option to hide the floating dictation pill when you don't need it.
Simplified Processing
Streamlined to two modes: Raw and Clean. Removed AI-dependent modes for faster, more predictable results.
YouTube URL Transcription
Paste a YouTube link, get a transcript. Powered by yt-dlp with automatic audio extraction.
7 Export Formats
Export transcripts as TXT, Markdown, SRT, VTT, DOCX, PDF, or JSON.
Clean Text Pipeline
Automatic filler word removal, custom vocabulary, and text snippets for polished transcriptions.
Custom Words & Snippets
Teach MacParakeet your jargon and create text expansion shortcuts.
In-App Feedback
Report bugs and request features directly from the app.
MacParakeet Launch · v0.1
First public release. System-wide dictation and file transcription powered by Parakeet TDT on Apple Silicon. 100% local, no cloud, no subscription.
January 2026
Project Initialized
Repo created. Research, spec, and scaffolding. Public app and website followed in February.