Release notes

Switching between Settings tabs is faster and smoother.

Wording across the app is clearer and more consistent.

Live captions now keep running when you switch headphones or audio devices.

Captions move around less while text is being revised, and the default window shows two lines of each language.

Translated captions are now slightly dimmer than the original text, making the two easier to tell apart.

Reduced energy use during Soniox live transcription.

Live Soniox transcription now skips stale audio when recognition falls too far behind, then resumes near the current audio.

Limited startup and reconnection buffering to the latest two seconds of live audio.

Added a notice when live audio is skipped, while preserving confirmed captions and ongoing audio recording.

File transcription continues to preserve audio for complete processing.

Reduced subtitle layout shifts from late punctuation, timing, and speaker updates.

Removed automatic scrolling animations so new subtitle lines appear in place.

Added automatic Soniox recovery when recognition falls behind, with limited retries and replay of unconfirmed audio.

Improved error messages for interrupted recognition by preserving the original failure.

Improved subtitle interaction and refined the settings interface.

Standardized Recognition Engine and Identify Speakers labels across Settings and the menu bar.

Improved live subtitle and translation stability during continuous speech.

Fixed translated speaker colors flickering while captions update.

Refined Soniox language settings and menu bar controls.

Added optional Soniox cloud speech recognition with automatic language detection and speaker identification. Selected audio is sent to Soniox when this engine is enabled.

Remembered language and speaker settings separately for each recognition engine, with clear checks for unsupported selections.

Improved long-running recognition with parallel connection renewal, more consistent speaker colors, and safer handling of overlapping subtitle timestamps.

Reused completed sentence translations with Apple and Google translation to reduce repeated requests and subtitle changes.

Fixed audio processing stalls caused by small capture buffers and repeated diagnostic work.

Made manual stop cancel pending recognition immediately, and improved final audio completion and error reporting.

Increased automatic speech recognition from four to eight candidate languages.

Skipped translation for captions already in the target language, avoiding duplicate bilingual subtitles while preserving Simplified and Traditional Chinese conversion.

Same-language captions remain visible in translation-only mode.

Prevented outdated translations from returning after language or translation-target changes.

Added more space around the subtitle preview in Settings.

Renamed the Chinese speech recognition option to Mandarin to distinguish it from Cantonese.

Fixed text overflowing shortcut status notifications during size transitions.

Prevented overlapping text when switching rapidly between short and detailed status messages, while keeping smooth resizing.

Aligned the automatic update check switch with the other switches in Settings.

Added customizable global shortcuts to toggle transcription and translation, cycle subtitle display, and restore the subtitle position.

Added a Shortcuts page in Settings. All shortcuts are off until you assign them, with checks for duplicate and conflicting key combinations.

Assigned shortcuts now appear beside their commands in the menu bar menu.

Added compact status notifications at the top of the screen, with smooth transitions and text-sized layouts to show operation progress and results.

Moved optional LID assistance to experimental builds and removed its model from the standard app for a smaller download. Automatic language selection through Apple Speech remains available.

Refined caption background spacing and kept wrapped lines connected.

Kept caption side and bottom edges consistent when expanding the window.

Improved close button alignment and its clickable area.

Fixed clipped text when scrolling selected captions, while preserving the selected text.

Reduced redundant caption updates and translation requests.

Improved memory management during transcription.

Made Check for Updates easier to find in the menu.

Added Developer ID code signing for macOS downloads.

Live captions for system audio, a selected application, or your microphone.

Translate speech and display original and translated captions together.

Adjust the caption window, text size, and appearance to fit your workspace.