This update fixes cases where short replies could disappear from transcripts and improves recording and playback reliability. • More consistent speaker corrections, renaming and Forget behavior. • Better cancellation and recovery for local models, plus automatic-tag retry handling. • Review notices and a retry option for transcripts with rejected timing or uncertain model annotations. • Clearer search, import and sharing guidance, with improved text contrast and large-text accessibility.

A new workspace for your conversations. • Cleaner recording, reading and source-linked Chat on iPhone and iPad. • Continuous Apple Watch recording, paired-iPhone processing and durable transfer recovery. • Included Small multilingual, optional Large multilingual, and restored Distil English-only transcription with accurate model status and recovery options. • Local Apple Intelligence or optional downloadable Qwen3.5-4B. • Readable speaker labels, preserved corrections, and bounded automatic-tag recovery. • Library and conversation Chat scopes, inspectable sources and action previews. • Optional consent-based cloud Chat using your own OpenAI, Anthropic or OpenRouter key.

Fixed Watch transfer queues that could get stuck after an app update or restart. Improved recovery of pending recordings and clarified transfer-queue errors.

Added Sync Now in Apple Watch Settings so you can manually retry recordings waiting to reach your iPhone. Improved retry handling and refreshed transfer status. Recordings remain on your Watch until your iPhone confirms safe receipt.

Stenographer 2.0.1 focuses on more reliable Apple Watch syncing and clearer transcripts. • More reliable Watch recording transfers after updates, restarts, reconnects, and interruptions. • Fixed a transfer queue stall that could require reopening the Watch app to send the next recording. • Fixed an extra empty recording that could appear after stopping a Watch recording. • Improved recording recovery and ensured transferred audio is added to the iPhone library before Watch delivery is confirmed. • See how many Watch recordings are waiting to transfer, with guidance for keeping the Watch app visible while syncing. • Fixed unrelated recordings being combined into one speaker turn just because both used a label such as “Speaker 1.” • Refined saved-speaker assignment handling and filtering of stray transcript text. • Fixed a timeline-loading crash affecting large recording libraries and improved Keep Screen On behavior during processing. Audio and transcripts continue to be processed privately on your devices.

Stenographer 2.0 makes every conversation easier to follow. • Identify who said what with private, on-device speaker diarization. • Name speakers and let on-device voice profiles suggest those names in other recordings when there’s a confident match. • Re-identify speakers in individual recordings or process older recordings without losing existing summaries and labels. • Read natural speaker turns instead of fragmented model segments. • Meet Seamless Transcript, a continuous, newspaper-like view that hides recording splits and processing chunks. • Prefer the old layout? The classic per-recording view remains available in Settings. • Add optional timestamps to transcripts. • Generate summaries and labels with Apple’s on-device Foundation Models on supported devices, with an optional downloaded local fallback. • Read plain transcripts sooner while speaker identification finishes in the background. • Search transcripts, summaries, and labels, then select and share exactly the text you need. • Handle large transcript and recording libraries with smoother pagination. • See clearer transcription, speaker-identification, summary, and iPhone processing status. • Keep finished Apple Watch recordings queued until iPhone safely receives them, with retries after offline periods, transfer failures, and relaunches. • Get more reliable first-use transcription and recording controls, plus improved launch stability for large or actively updating libraries. Audio and transcripts stay on your devices. Models may download when needed, but your recordings are not uploaded for processing.

Adds audio file import, improves Watch recording reliability and complication start/stop behavior, updates the app icon, and bundles on-device model assets for a smoother offline experience.

Bug and stability fixes.

Set optional max recording time limit on iOS or watchOS.

Better help section.

Watch complication to start and stop recording from the face.

New Qwen3 model for improved local LLM processing. Ability to label transcriptions for easier lookup. Auto generate labels with local LLM.

Add multi-selection to merge or delete recordings.

Fix audio conflicts.

Ensure transcriber cleanup

Right-size recording permissions.

Handle audio interrupts. Fix recording restart bug.

Multilingual transcription support. Translation to English. Long press to copy text. Customize UI ordering. Cleaner concurrency.

Fix processing empty files

Expose LLM prompt for summarization. Minor UI fixes.

Hallucination detection. Improve background processing.

Swipable actions in list view. Help/Welcome menu.

Add welcome screen and help menu. Repair minor search UI issues.

Process multiple recordings in parallel. Snappier UI.

Repair background processing. Highlight text on search.