DictaFlow release notes

Final microphone-recovery and recording reliability improvements

• Reliable default-microphone recovery: DictaFlow refreshes the Windows audio runtime once when the default input temporarily disappears, while preserving explicit microphone choices.
• Safe bounded diagnostics: a slow Windows device query can no longer retain the shared audio lock and interfere with later recordings.
• Stress-tested recording lifecycle: rapid recording controls, microphone tests, session changes, and repeated stream opens now share stronger ownership and cleanup safeguards.

More responsive recording and stronger microphone recovery

• More responsive recording controls: hotkey, mouse, foot-pedal, remote-session, and in-app recording commands now run through a resilient serialized worker.
• Better Windows microphone recovery: when the Windows default microphone temporarily disappears, DictaFlow refreshes the audio-device runtime once and retries without silently switching to another microphone.
• Clearer microphone guidance: persistent default- or selected-device failures now provide actionable Audio Input instructions instead of exposing a misleading system error.
• Stronger long-session reliability: audio ownership and cleanup are more robust across startup, microphone tests, lock and unlock, sleep and wake, and remote-session reconnects.

AI Mode shortcut reliability on Mac

• AI Mode no longer ends immediately: fixed an issue that could make the Control+Fn shortcut briefly appear and stop while DictaFlow prepared selected text.
• Correct shortcut release handling: background selected-text capture can no longer imitate a shortcut release; recording continues until a required shortcut key is actually released.
• Mac speed improvements retained: the faster selected-text preparation, lossless audio upload, and connection prewarming remain enabled.

Italian, Dutch, and Traditional Chinese on Mac

• Three complete Mac interface localizations: DictaFlow and DictaFlow Medical now support Italian, Dutch, and Traditional Chinese throughout sign-in, onboarding, the main app, and Settings.
• Natural, context-aware wording: every new translation was manually written and reviewed for the screen and workflow where it appears.
• Localized clinical workflows: DictaFlow Medical includes translated clinical dictation, consultation-note, and Ambient Scribe experiences in all three languages.
• Smarter first-run defaults: new users receive an appropriate dictation-language recommendation based on their selected interface language.

More languages, faster dictation, and stronger Windows reliability

• Balanced Home layout: Recent Activity once again keeps the wider, easier-to-scan area while Transcript Preview stays compact.
• Eight interface languages: Windows Consumer and Medical now include complete offline interface localization in English, French, German, Mexican Spanish, Brazilian Portuguese, Japanese, Korean, and Simplified Chinese.
• Choose your interface before signing in: a language selector is now available on the account screen and again in Settings, with changes applied immediately.
• A much broader dictation-language picker: commonly used languages appear first, followed by an alphabetized list covering more than 100 language and regional options supported by Supersonic Sprint.
• Automatic model compatibility: when a fixed language is not supported by Studio Session, DictaFlow switches new selections to Supersonic Sprint. Settings also explains the compatibility issue if Studio Session is selected later; Supersonic Sprint and Command Suite remain available.
• Faster recording handoff: optimized lossless audio uploads reduce work after recording and help dictation reach the cursor sooner.
• More dependable hotkeys: recording controls recover more reliably after lock and unlock, sleep and wake, and remote-session reconnects.
• Faster cloud dictation: speed and stability improvements across Studio Session and Supersonic Sprint.

Lower latency and more resilient desktop sessions

• Reduced Windows dictation latency and tightened the path from releasing the shortcut to seeing text.
• Improved hook recovery after Windows session changes, resumes, and reconnects.
• Made startup prewarming, backend failover, and long-running account sessions more resilient.

A clearer first-run experience across platforms

• Introduced account-first onboarding and a compact Quick Start for Windows, Mac, and iOS.
• Added tailored first-run experiences for DictaFlow and DictaFlow Medical, including microphone setup, shortcut guidance, and a first practice dictation.
• Hardened audio, recording, and network lifecycles to prevent hangs and recover cleanly when native operations time out.
• Improved preview transcription quality and provider fallback behavior.

Better formatting and safer releases

• Added automatic email formatting for Local Offline transcription.
• Improved packaging checks and release tooling for more dependable desktop builds.

Local Offline expanded into AI Mode

• Made Local Offline transcription available in AI Mode.
• Improved local-model download, management, and persistence across restarts.

On-device dictation

• Introduced Local Offline transcription on Windows with an on-device speech model.
• Expanded local options and offline fallbacks on Apple platforms.

Safer edits and better mobile dictation

• Made AI Smart Edit more conservative by default so it preserves intent instead of inventing content.
• Improved number formatting in automatic language detection.
• Improved iOS keyboard controls and dictation behavior.

Smoother sign-in and onboarding

• Added long-lived desktop sessions so users stay signed in more reliably.
• Refined iOS onboarding to get users to a successful first dictation sooner.

Quick Start on Mac and stronger service fallback

• Added the guided Quick Start experience on macOS.
• Improved service failover and transient retry handling.

Availability and preview safeguards

• Expanded backend failover so dictation can recover from a temporarily unavailable service.
• Added safeguards for first-use preview dictations and provider errors.

A faster path to the first Windows dictation

• Introduced a conversion-focused Windows preview and onboarding flow.
• Improved the reliability and safety of anonymous preview transcription.

More sign-in choices and better support diagnostics

• Added Sign in with Apple on Windows.
• Improved diagnostics for support, recording-trigger controls, output behavior, and email formatting.

Earlier highlights

• Added Ambient Scribe, speaker labels, clinical templates, longer Medical recordings, and EHR-friendly text handoff.
• Expanded Knowledge Base imports, selection-aware replacement, strict verbatim controls, mouse and foot-pedal triggers, and multilingual dictation.

These notes highlight user-facing changes and are not an exhaustive engineering changelog. New releases will be added here at the top.