DictationBridge Lite 0.3.0 beta 1

DictationBridge Lite 0.3.0 beta 1 improves feedback from both Windows dictation systems in current NVDA.

What changed

• Online Windows+H dictation now announces finalized phrases instead of provisional word fragments and rapid repeats.
• Rotating online tips and routine Listening or Thinking messages are suppressed while genuine errors remain available.
• The tooltip-race repair is preserved, so a command or suggestion tooltip does not silence the following dictated text.
• Legacy offline Windows Speech Recognition dictated-text feedback remains responsive.
• A new NVDA menu > Tools > DictationBridge Lite submenu opens online and offline Windows speech settings and includes a checked Speak dictated text toggle.
• Public and bundled documentation now explain which original DictationBridge functions were preserved, replaced, partially preserved, or removed.

Validation

• 29 automated regression tests passed.
• Fresh Win32 and x64 native components were built by GitHub Actions.
• Online and offline dictation behavior was tested with NVDA on Windows 10.

Known beta limitations

• Online speech is delayed until Windows finalizes the phrase.
• Legacy offline correction panels are announced only when Windows exposes a usable accessibility event, so automatic panel speech remains inconsistent.
• Windows 11 and a broader application matrix still need community testing.

The only attached release asset is the installable NVDA add-on. GitHub provides the corresponding source archives automatically, and the complete project remains cloneable and modifiable under its included licenses.

SHA-256: b1ef76e38748c83ab45a8b681bdceaf623e6e2d6f400b89a86c3fb082cf5d357

Changes since the previous beta: v0.2.1-beta.1...v0.3.0-beta.1

DictationBridge Lite v0.2.1-beta.1

DictationBridge Lite 0.2.1 beta modernizes the dictated-text feedback from the original DictationBridge for current NVDA.

Added or repaired

• Keeps both a transient command tooltip and the following dictated text audible.
• Restores legacy offline Windows Speech Recognition text echo on current NVDA.
• Restores automatic reading of legacy WSR alternates and spelling-panel choices.
• Supports current 32-bit and 64-bit NVDA and observes 32-bit and 64-bit applications.
• Deduplicates overlapping native and NVDA insertion or deletion events.

Preserved

• Short-delay dictated-text feedback, new-line and new-paragraph announcements.
• Deleted and replacement-text feedback.
• Accessible Windows Speech Recognition training passages and the reread command.

Intentionally omitted

• Dragon-specific integration and microphone announcements.
• JAWS packaging and legacy screen-reader or WSR command macros.
• Microsoft Word-specific automation; Lite uses generic Text Services Framework observation instead.

Known beta bug

During online Windows+H dictation, NVDA can announce incomplete word fragments while Windows is updating the text and then announce the completed word or phrase. The final result is still spoken, but the preliminary fragments can be noisy.

The release has one installable asset. The complete corresponding source can be cloned from the repository or downloaded through GitHub's automatic source archives.