Changelog

Track every update, feature, and improvement to Transkript

Version 4.2 June 2026
FluidAudio 0.15.5 Dependency refresh Release metadata

FluidAudio Update

  • FluidAudio: Updated to 0.15.5 for the 4.2 release.
  • Existing workflows: Speaker recognition, diarization presets, and private on-device processing remain unchanged.
  • Foundation: This dependency refresh prepares the app baseline for newer upstream FluidAudio on-device ASR work without exposing those engines as separate user-facing options yet.
  • Release metadata: App version, build number, website, and App Store metadata are refreshed for 4.2.
Version 4.1 May 2026
Share import fixes Cancellation cleanup Safer exports Dependency updates

Reliability Fixes

  • Sharing: Voice Memos and share-extension imports now use unique inbox filenames and a pending import queue.
  • Transcription: Stopping a task now prevents late transcription results from overwriting the session state.
  • Summaries and exports: Local-model thinking blocks, edited transcript word counts, and WebVTT escaping are handled more safely.
  • Dependencies: FluidAudio is updated to 0.14.7 and MLX Swift LM to 3.31.3 for the 4.1 release.
Version 4.0 May 2026
Qwen 3 on iPad/Mac Gemma 4 Preview on Mac FluidAudio 0.14.x Local AI safeguards

What's New

  • Qwen 3: Optional experimental local AI backend on iPad with the smaller 1.7B model and on Mac-class devices. On iPhone, Apple Intelligence remains the stable default.
  • Gemma 4 Preview: Optional experimental local AI backend for Mac-class devices. It is clearly labeled as Preview in the app.
  • Conservative context handling: Long transcripts are chunked with an 8K safe context for Gemma 4 and compact safeguards for Qwen 3 on supported iPad/Mac devices.
  • Clearer transcription progress: FluidAudio 0.14.x uses native progress callbacks and a simpler 3-step display for long speaker-recognition transcriptions.
  • Availability checks: Local AI backends only appear after platform, model download, RAM, and Metal checks pass.
Version 3.10 May 2026
Qwen 3 summaries FluidAudio 0.14.5 Progress & reliability

What's New

  • Qwen 3 AI backend: Optional local summaries and translations with a 32K-token context for longer transcripts. The model runs on-device via MLX and requires a separate download.
  • Smarter long-summary flow: Long transcripts are split into chunks, summarized section by section, then merged into a final summary with visible progress.
  • FluidAudio 0.14.5: Updated speaker diarization with improved reliability and clearer progress states for long audio files.
  • Release-ready safeguards: Qwen 3 now only appears as available after the model is fully downloaded, with clearer RAM, Metal GPU, and download messages.
Version 3.9 April 2026
FluidAudio 0.13.4 Smart Model Selection Bug Fixes

What's New

  • FluidAudio 0.13.4: Major SDK update — dependencies reduced from 12 packages to just 1, resulting in a leaner app and faster build times.
  • Smart Model Selection: Transkript now automatically picks the right speech model for your device. Under 8 GB RAM gets the lightweight 110M model (~600 MB, English); 8 GB and above gets the full v3 model for all 25 European languages (~1.5 GB).
  • Language Transparency: When the 110M model is downloaded, the language picker correctly limits FluidAudio to English only. A clear note in Settings explains the limitation and offers a one-tap option to switch to v3.

Bug Fixes

  • Apple Intelligence Detection: Fixed "Model assets unavailable" error on devices without Apple Intelligence hardware (e.g. iPad 10th gen). The app now correctly checks hardware eligibility before attempting summarisation or translation.
  • Delete Models UI: Deleting FluidAudio models now instantly reflects in Settings — the previous "ready" state no longer persists after deletion.
  • Long Recordings: Added a progress hint for recordings over 20 minutes with Apple Intelligence, so it's clear the app is still working and hasn't stalled.
  • Summarise Button: The summarise button is now correctly disabled on devices without Apple Intelligence, rather than showing a confusing error after tapping.
Version 3.8 March 2026
FluidAudio 0.12.3 Bug Fixes Accessibility

Bug Fixes

  • FluidAudio 0.12.3: Updated to the latest engine with task cancellation support and improved download progress tracking.
  • Memory Leak Fixed: Diarization models (~300 MB) are now correctly released after plain transcription runs, reducing memory pressure on iPhone.
  • Enhancement State Fix: Audio enhancement UI no longer gets stuck in a loading state when enhancement fails.
  • Cancellation Handling: Cancelling a transcription no longer shows an error dialog — the session silently reverts to its previous state.
  • Model Reload Progress: After an app restart, reloading FluidAudio models now shows progress in the UI instead of a silent freeze.
  • Recording Resume: Fixed a case where resuming a paused recording could silently fail while the UI showed it as active.

Improvements

  • Dynamic Type: Recording timer, tab bar icons, and several other elements now scale correctly with system font size settings.
  • VoiceOver: Action bar buttons on iPhone now have proper accessibility labels — VoiceOver users can identify all actions correctly.
  • File Protection: Session transcripts and audio files are now protected with iOS complete file protection, encrypted when the device is locked.
  • Localization: Fixed status messages during model download that were incorrectly shown in German to English, French, and Dutch users.
Version 3.7 February 2026
FluidAudio 0.12 iPhone Reliability Liquid Glass Accessibility

What's New

  • FluidAudio 0.12 Engine: Major upgrade to the speaker recognition engine with streaming audio processing for better memory efficiency and improved speaker identification.
  • iPhone Reliability: Speaker recognition now works reliably on iPhone. Fixed an issue where transcription with speakers could produce empty results on memory-constrained devices.
  • Orphaned Word Recovery: Words near speaker segment boundaries are no longer lost — they're automatically assigned to the nearest speaker.
  • iOS 26 Liquid Glass: Completely redesigned with Apple's new Liquid Glass design language. Glass tab bar, smooth morphing animations, and subtle visual depth throughout.

Improvements

  • Memory Optimization: Chunked audio processing (10-second segments) prevents out-of-memory on long recordings. ML models (~1.5 GB) are released from memory after use.
  • Keyboard Shortcuts: ⌘, for Settings, ⌘I to Import, ⌘R to Record — perfect for iPad with keyboard and Mac.
  • iPad & Mac Layout: Streamlined toolbar removes duplicate settings button. Cleaner split-view experience.
  • Accessibility: Full VoiceOver support with labeled buttons and hidden decorative elements. Dynamic Type scales all text with your preferred size. Semantic heading structure in summaries for screen readers.
  • Read Aloud: Listen to AI summaries with text-to-speech. Automatic language and voice selection. Pause, resume, and stop controls.
Version 3.6 December 2025
Speaker Editing Audio Enhancement Meeting Summaries

What's New

  • Speaker Name Editing: Finally! Rename speakers directly in the transcript. Tap on "Speaker A" and change it to "John". The name is applied to all segments from that speaker.
  • Audio Enhancement: New preprocessing feature to improve audio quality before transcription. Applies high-pass filter (removes rumble), presence boost (clearer speech), soft compression, and volume normalization.
  • Improved Meeting Summaries: Meeting-style summaries now specifically focus on extracting decisions, agreements, and action items (TODOs) with responsible persons.

Improvements

  • Real-time UI Updates: Summaries and translations now appear immediately after generation, no need to navigate away and back.
  • Better Progress Feedback: Audio enhancement shows animated progress with percentage indicator.
  • Various UI polish and bug fixes
Version 3.5 December 2025
Session Library 4-Tab Interface Video Preview

What's New

  • Session-Based Architecture: Completely redesigned around sessions. Each recording becomes a session containing transcript, translations, summaries, and subtitles – all in one place.
  • Library-First Design: The library is now your home screen. See all sessions at a glance with status badges, quick actions, and floating buttons for recording and import.
  • 4-Tab Interface: Navigate easily between Transcript, Translations, Summaries, and Subtitles tabs with iOS 26 Liquid Glass design.
  • Video Subtitle Preview: Preview your subtitles directly on the video with live sync. Switch between original, optimized, and translated subtitles.
  • Custom Storage: Store your sessions in iCloud, Proton Drive, or any folder. Folder migration dialog helps you move or start fresh when changing locations.
  • Auto-Sync: Pull-to-refresh and automatic refresh when app becomes active. Perfect for shared folders across devices.
  • Drag & Drop: Drop audio and video files directly into the library on Mac.

Improvements

  • Multiple Translations: Store unlimited translations per session. Each in a different language if you want.
  • Multiple Summaries: Create summaries in different styles (brief, medium, detailed) and languages.
  • iOS 26 Liquid Glass Tabs: Beautiful iOS 26 design with color-coded icons. Icon-only on iPhone, icon + text on iPad and Mac.
  • Location-Based Naming: New recordings are automatically named after your current location.
Version 3.0 December 2025
Speaker Recognition Voice Memos Share 100% Apple Intelligence

What's New

  • Speaker Recognition (Diarization): Automatically identify different speakers in your recordings. Perfect for meetings, interviews, and podcasts. Color-coded labels show who said what.
  • Voice Memos Integration: Share recordings directly from Apple's Voice Memos app to Transkript via the Share Sheet.
  • FluidAudio Engine: New on-device speaker diarization powered by FluidAudio SDK for accurate speaker identification.
  • Speaker Presets: Choose from Interview (2-3 speakers), Meeting (4-6 speakers), Conversation (2 speakers), or Custom settings.

Architecture Changes

  • Whisper Replaced by FluidAudio: Whisper has been replaced by FluidAudio SDK which provides both speech recognition (25 European languages) and speaker diarization. Apple Intelligence remains available for transcription-only workflows.
  • iOS 26+ Only: Requires iOS 26.1 or later with Apple Intelligence enabled.
  • Share Extension: New Share Extension allows receiving audio files from any app via the Share Sheet.
Version 2.6 December 2025
Word Export Long Text Support

What's New

  • DOCX/Word Export: Export your summaries as Microsoft Word documents (.docx) in addition to PDF. Perfect for editing and sharing.
  • Hierarchical Summarization: Long transcripts are now automatically split into chunks, summarized individually, then combined into a final summary. No more hanging on very long texts.

Improvements

  • Accurate Progress: Progress indicator now based on actual audio position instead of text length.
  • Better Time Estimates: Shows estimated time remaining during transcription.
  • Cross-platform ZIP implementation for reliable DOCX creation
Version 2.5 December 2025
Simplified UI Auto Engine Selection

What's New

  • Automatic Engine Selection: The app now automatically chooses the best AI engine (Apple Intelligence or Whisper) based on your selected language.
  • Simplified Language Picker: See availability status for each language with green bolt (Apple Intelligence) or orange checkmark (Whisper).
  • One-Click Whisper Download: Easy prompt to download Whisper model for 90+ additional languages.
  • Streamlined Settings: Removed manual engine picker for a cleaner, more intuitive interface.

Improvements

  • Dynamic language refresh when Whisper models are downloaded
  • Better visual indicators for engine availability
  • Improved first-time user experience
Version 2.0 December 2025
Whisper AI Subtitle Optimizer Video Player

New Features

  • Whisper AI Integration: OpenAI's Whisper as an alternative transcription engine supporting 99 languages with high accuracy.
  • Dual Engine Choice: Switch between Apple Intelligence (live transcription) and Whisper (batch processing) on the main screen.
  • Subtitle Optimizer: Automatically split long segments into readable subtitles with customizable max characters, duration, and timing.
  • Show Original Text: View the original text for incompletely translated segments to understand context.
  • Whisper Processing Animation: Beautiful animated waveform display while Whisper processes audio.

Improvements

  • Quick Start: Option to automatically start transcription when dropping a file (configurable in Settings).
  • Animated Progress: Continuously animated progress ring and waveform icon during transcription for better visual feedback.
  • App Header Bar: New header with app icon, name, and menu for quick access to Settings and About.
  • Button Labels: All action buttons now show text labels for better clarity and usability.
  • Full-Screen Export: Export dialogs now use full screen for better overview of options.
  • Video Player Controls: Playback speed (0.5x-2x), AirPlay, video fill toggle, and native macOS fullscreen.
  • Audio Player Speed: Variable playback speed for audio preview (0.5x to 2x).
  • Improved Translation View: Preview button moved to action bar, consistent with transcript view.
  • Smart Live Preview: Live transcription preview only shown for Apple Intelligence (not Whisper).

Whisper Languages

  • 99 Languages: Afrikaans, Albanian, Amharic, Arabic, Armenian, Azerbaijani, Basque, Belarusian, Bengali, Bosnian, Bulgarian, Burmese, Catalan, Chinese, Croatian, Czech, Danish, Dutch, English, Estonian, Finnish, French, Galician, Georgian, German, Greek, Gujarati, Hebrew, Hindi, Hungarian, Icelandic, Indonesian, Italian, Japanese, Kannada, Kazakh, Khmer, Korean, Lao, Latvian, Lithuanian, Macedonian, Malay, Malayalam, Maltese, Maori, Marathi, Mongolian, Nepali, Norwegian, Pashto, Persian, Polish, Portuguese, Punjabi, Romanian, Russian, Serbian, Sinhala, Slovak, Slovenian, Somali, Spanish, Swahili, Swedish, Tagalog, Tamil, Tatar, Telugu, Thai, Tibetan, Turkish, Turkmen, Ukrainian, Urdu, Uzbek, Vietnamese, Welsh, Yiddish, Yoruba, and more.
Version 1.1 December 2025
Video Support VTT Export Subtitle Preview

New Features

  • Video to Subtitles: Load MP4, MOV, or M4V video files directly. Audio is automatically extracted for transcription.
  • WebVTT Export: New VTT format for web video players alongside existing SRT support.
  • Video Thumbnail Preview: See a preview frame from your video in the file info card.
  • Subtitle Sync Preview: Preview your subtitles overlaid on the original video to verify timing before export.
  • Multilingual Subtitles: Create subtitles in the original language AND translated versions in 34 languages.

Improvements

  • Enhanced file type detection for video formats
  • Visual distinction between audio and video files in UI
  • Improved export dialog with format-specific hints
Version 1.0 November 2025
Initial Release

Core Features

  • Audio Transcription: Support for MP3, M4A, WAV, AAC, AIFF, and CAF files with word-level timestamps.
  • Live Recording: Record audio directly in-app with real-time transcription preview.
  • AI Summarization: Generate intelligent summaries using Apple Intelligence in multiple languages.
  • Translation: Translate transcripts to 34 languages while preserving timestamps for subtitle export.
  • Multiple Export Formats: TXT, SRT subtitles, JSON, and PDF with formatted summaries.
  • Library Management: Save, organize, and search your transcriptions with iCloud sync.
  • Siri & Shortcuts: Voice commands and automation via the Shortcuts app.
  • Control Center Widgets: Quick-launch buttons for iOS Control Center.
  • Live Activities: Real-time progress on Lock Screen and Dynamic Island.
  • Audio Preview Player: Listen to audio files before transcription with playback controls.
  • 100% Privacy: All processing on-device with Apple Intelligence. No cloud uploads.

Supported Languages

  • Transcription: English, German, French, Spanish, Italian, Dutch, Portuguese, Japanese, Korean, Chinese, and more
  • Translation: 34 target languages including Arabic, Hebrew, Thai, Vietnamese, Indonesian, Turkish, Greek, Russian, Ukrainian
  • UI Languages: English, German, French, Dutch

Try the Latest Version

Try Transkript 4.2 with FluidAudio 0.15.5, Qwen 3 on iPad/Mac, and Gemma 4 Preview on Mac-class devices.

Download on the App Store