Téléchargements et notes de version

Parrot Scribe 0.6.0 improves the Library with better folder organization and richer session details, making larger collections easier to manage.

What's New

• Folders: Added Library folder foundations so sessions are easier to organize as your archive grows.
• Session Details: Added richer metadata in Library to make recordings easier to identify at a glance.
• Library Workflow: Improved folder interactions and polish so organizing and browsing feels smoother and more reliable.

---

Parrot Scribe 0.5.0 expands Library import so you can bring in more recordings, including video files with audio, with better progress handling and cleanup.

What's New

• Video Import: Import video files with audio tracks directly into Library for transcription.
• Import Reliability: Improved cleanup and recovery when an import is cancelled or a file fails partway through.
• Progress & Control: Improved import progress behavior across Library and menu bar surfaces so long-running imports are easier to follow.

---

Parrot Scribe 0.4.0 is the cumulative stable release from the 0.3.2 line, including all major improvements from the 0.4.0 alpha, beta, and release-candidate cycle.

Important Before You Upgrade

• Pre-0.4.0 sessions are not compatible with 0.4.0+.
• There is no migration path in 0.4.0+.
• Export legacy sessions with 0.3.2 before upgrading.
• Data directory paths:
• pre-0.4.0: ~/Library/Application Support/ParrotScribe/
• 0.4.0+: ~/Library/Application Support/Parrot Scribe/
• pscribe transition: Most pscribe functionality is now built directly into Parrot Scribe's integrated MCP server, so MCP no longer depends on the standalone pscribe tool. Non-MCP pscribe commands are expected to return in a future update.

What's New Since 0.3.2

• Storage v2 Foundation: Introduced a new storage foundation for sessions, comments, and encrypted data handling to support long-term reliability.
• Built-in MCP Transition: MCP is now built into Parrot Scribe with updated token-based access and setup flow in Settings.
• Audio Import: Added audio file import workflows so you can bring existing recordings into Library for transcription.
• Session Comments: Added session comments with add, edit, and clear actions in Library.

Reliability Improvements

• Library & Session Reliability: Improved Library and session consistency, startup behavior, metadata clarity, and review flow.
• Speaker Management Improvements: Improved speaker merge, removal, bulk actions, warnings, and known-speaker persistence behavior.
• Installer Reliability Fixes: Improved upgrade/install behavior to reduce failed updates and stale legacy artifacts.

---

What's Fixed

• Session Viewer & Library: Fixed an error when opening encrypted transcripts in the Session Viewer and Library windows ("Security manager is required").
• Build Reliability: The security manager dependency is now enforced at compile time, preventing similar issues in the future.

What's Fixed

• Transcript Migration: Fixed an issue where transcript encryption migration failed on finalized session files, causing a migration error on every app launch.

ParrotScribe 0.3.0 introduces Speaker Recognition, encrypted storage, and a new Library window for managing your transcription sessions.

What's New

• Speaker Recognition: ParrotScribe can now identify who's speaking in your transcripts. After a session, name your speakers and the app will recognize them in future sessions. Speaker data is protected with biometric authentication and encrypted on-device.
• Library Window: A new Library window (⌃⌘L) for browsing all your sessions and managing enrolled speakers. Includes a Session Viewer with synchronized audio playback and waveform visualization.
• Audio Retention: Optionally retain encrypted session audio for playback and speaker identification. Configure retention period or storage limits in Settings.
• Encrypted Storage: All transcripts, speaker data, and retained audio are now encrypted at rest for an additional layer of privacy.
• Export Command: Use pscribe export to export transcripts from the command line.
• Keyboard Shortcuts: Configure system-wide hotkeys to toggle, start, or stop transcription — all without switching apps. Full keyboard navigation throughout the app.
• Notifications: Get notified when transcription starts, stops, or begins a new session.
• Accessibility: Full VoiceOver support, Reduce Motion support, and text labels alongside all color indicators.
• Extended Trial: Free trial period extended from 3 to 7 days.

What's Improved

• Speaker Identification: Clearer visual feedback when identifying speakers, with accent-colored toolbar buttons and notification-based attribution.
• CLI: Speaker names now appear correctly in cat, tail, and grep output. The cat command defaults to showing the last session.
• System Audio: Improved system audio capture using ScreenCaptureKit for better reliability.
• Data Storage: Transcripts are now stored in a protected location with automatic migration from earlier versions.
• Uninstall Safety: pscribe uninstall now preserves your transcripts and speaker data by default.

What's Fixed

• Upgrade Reliability: All existing sessions are preserved correctly when upgrading from earlier versions.
• Window Management: Pressing ⌘Q now closes windows and returns to menu-bar mode instead of quitting the app.
• Transcription Quality: Improved accuracy by filtering AI model artifacts from output.

---

Reliability improvements for the 0.2.x release series.

What's Fixed

• Startup Reliability: Fixed an issue where CLI commands (like pscribe status) might not work after the app started automatically at login.
• Recording Reliability: Fixed an issue where the app could appear to be listening but not actually record audio.
• Settings: The output directory selector now uses the standard macOS file picker for a cleaner experience.

---

New CLI commands for searching and viewing your transcription history.

What's New

• Grep Command: Search across all your transcription sessions with pscribe grep. Find specific words or phrases with full regex support, context lines (-A, -B, -C), and session filtering.
• Cat Command: Display complete transcription sessions with pscribe cat. View entire sessions or filter by status.
• Improved Polling: The MCP integration now supports efficient offset-based polling for real-time transcript updates.

---

Fixed an issue with the "Launch at Login" setting.

What's Fixed

• Launch at Login: The toggle in Settings now correctly registers ParrotScribe as a login item. Previously, the preference was saved but the app wouldn't actually start at login.

---

ParrotScribe is a native macOS menu bar app that transcribes audio in real-time using on-device AI. Your audio never leaves your Mac.

What's New in 0.1.0

Real-Time Transcription

• Capture system audio, microphone, or both simultaneously
• Powered by WhisperKit with Apple Neural Engine acceleration
• Works completely offline — no internet required, no data sent anywhere

Multi-Language Support

• Full interface localization: English, German, French, Spanish, and Swedish
• Speech recognition for 99+ languages via Whisper models
• Configure expected languages to improve transcription accuracy

Flexible Output Formats

• Export transcripts in 8 formats: TOON, CSV, JSON Lines, Markdown, SRT, VTT, Syslog, and Plain Text
• Use SRT/VTT for subtitles, CSV/JSONL for data analysis, or Markdown for notes
• Session-based files with automatic crash recovery

Smart Model Management

• Hardware-aware model recommendations based on your Mac's chip and RAM
• Download progress with SHA256 verification
• Automatic cleanup of old model versions

Command Line Power

• Stream live transcripts: pscribe tail -f
• View recent entries: pscribe tail -n 50
• Output in any format: pscribe tail -o json
• Full config sync between app and CLI

Automatic Updates

• Sparkle-powered updates with dual channels (stable/latest)
• Choose: notify only, download automatically, or install automatically

Simple Licensing

• 3-day free trial — no credit card required
• One-click activation from the website's thank-you page
• Works offline after activation

Requirements

• macOS 15.0 (Sequoia) or later
• Apple Silicon Mac (M1, M2, M3, M4, M5)
• Audio Input permission for microphone and system audio capture

---