v2.1.21

CodeSplash AI 2.1.21

CodeSplash AI 2.1.21 ships the first phase of in-app voice dictation. You can now speak into supported CodeSplash input surfaces, manage dictation from a new Voice settings page, and use a local on-device speech pipeline designed for desktop reliability and privacy-conscious workflows.

Features

• Phase 1 voice dictation is now available in-app so you can speak into CodeSplash instead of typing in supported surfaces.
• New Voice settings page with dictation controls, including microphone selection and polish model options.
• Dictation model setup UI to help configure the local speech models required for voice input.
• Broader in-app dictation targets across major text-entry surfaces, including AI composer experiences, terminal-related input flows, and focused in-app text fields.

Fixes

• More reliable packaged dictation support by verifying the required native speech runtime is included before desktop builds ship.
• Lower risk of first-run dictation failures in packaged apps through improved handling of platform-specific native speech dependencies.

Improvements

• Local speech recognition now runs off the main process, helping keep the desktop app responsive while dictation is active.
• Voice settings are now indexed in Settings search, making controls like microphone and polish model easier to find.
• Desktop packaging support was expanded for the speech runtime across supported release platforms.

Notes

• This release is the first shipped phase of dictation.
• The strongest support in this version is for in-app dictation targets inside CodeSplash.
• Speech recognition is positioned as local on-device transcription, not a cloud-audio workflow.

If voice input is part of your workflow, update to 2.1.21 and try dictation in your main CodeSplash AI surfaces.