1.3.2 beta

What's new

• On macOS 26.4 and later, on-device translation uses recent conversation context. As new speech arrives, recent translations may update while keeping their original speaker and timestamp.
• Saved recordings can be transcribed, continued, transcribed again with different languages, or translated again. Processing can pause and resume across app restarts, and new results can be reviewed before applying them.
• VoiceOver announcements, keyboard focus and navigation, visible status labels, and macOS text, contrast, motion, and transparency preferences are supported throughout core workflows.
• Audio capture updates avoid unnecessary interface invalidations, and accessibility announcement history stays bounded.
• Recording metadata validation and audio driver shared-state handling are more robust.

Contextual translation requires macOS 26.4 or later; earlier supported systems translate each entry individually.

Changes

• Fix Swift concurrency ownership in PCM conversion callbacks.
• Keep connection setup instructions readable on narrow screens.

1.3.1 beta

What's new

• On macOS 26.4 and later, on-device translation uses recent conversation context. As new speech arrives, recent translations may update while keeping their original speaker and timestamp.
• Saved recordings can be transcribed, continued, transcribed again with different languages, or translated again. Processing can pause and resume across app restarts, and new results can be reviewed before applying them.
• VoiceOver announcements, keyboard focus and navigation, visible status labels, and macOS text, contrast, motion, and transparency preferences are supported throughout core workflows.
• Audio capture updates avoid unnecessary interface invalidations, and accessibility announcement history stays bounded.
• Recording metadata validation and audio driver shared-state handling are more robust.

Contextual translation requires macOS 26.4 or later; earlier supported systems translate each entry individually.

1.1.0 (10)

• Connect selectable Agent and calling apps, with separate listening levels and optional built-in speaker fallback.
• Record, transcribe and translate sessions, with transcript search and automatic following of the latest text.
• Rename recordings, inspect missing-audio intervals and export complete sessions or individual audio tracks.
• Show the selected playback source and highlight only the Play segment of the split control.
• Update listening controls without repeated application discovery and show device error details in an expandable area.
• Support English and Korean interfaces with adjustable text size.

1.1.0 beta (7)

• Give the Play dropdown a flat inner highlight and a full click target.
• Recognize the current ChatGPT app and make Discord available in Caller choices.
• Return to Settings after closing Audio Setup opened from Settings.
• Align startup guidance with the audio controls.
• Keep transcript messages above the processing indicator while following the latest text.
• Add a recorded-call demo to the English and Korean READMEs.
• Clarify audio-provider and app-removal information.

1.1.0 beta (6)

Switchboard 1.1.0 beta (6), for Apple Silicon Macs running macOS 26 or later.

Actual-call verification has not been completed for this build.

Changes

• Keep the live transcript at the latest message as text and delayed translations arrive. Preserve the reading position when scrolling up, and resume following at the bottom.

• Fix an Xcode 26.0.1 compiler crash in the recording-list selection binding.

• Named sessions with explicit Start, Pause, Resume, and Save, plus independent recording and transcription controls.

• Selectable Agent and caller applications, separate waveforms and listening levels, and optional built-in speaker fallback.

• Live transcription and translation, transcript search with Command-F, and timestamp navigation.

• A resizable recording sidebar with session details, renaming, source-specific playback, and complete .switchboard exports. Existing .mihrecording files remain readable.

• English and Korean interfaces, adjustable text size, and clearer permission and error displays.

• New installations confirm their save folder, with ~/Documents/Switchboard suggested. Existing installations retain their saved settings and recording locations.

• Recording timing, capture ownership, recovery, and concurrency fixes.

Known issues in this beta

• The playback dropdown can still show a circular highlight instead of a flat inner edge.
• The application picker can misidentify ChatGPT as ChatGPT Classic, and may not show Discord after it launches.
• Closing Audio Setup opened from Settings can return to the main screen instead of Settings.
• The message above Start has an alignment issue.

Switchboard 1.0.1

• Fix virtual audio device lookup and property responses.
• Prevent recording startup and shutdown races, and preserve completed recording metadata.
• Keep audio setup and recording-library loading responsive.
• Improve recording search, keyboard focus, and audio setup controls.
• Add macOS builds and memory and thread checks to CI.

Initial Release

• Connect a voice Agent in Chrome with the macOS Phone app.
• Monitor caller and Agent audio with separate volume controls and live waveforms.
• Record conversations, play and seek recordings, and export mixed M4A or separate WAV files.
• Choose English or Korean on first launch or in Settings.