Release 1.4.0

1.4.0 (2026-02-08)

New

1. Speech.speak supports per‑utterance options via Speech.speak(text, options) and returns an utterance ID for tracking.

Breaking Changes

1. Speech.initialize → Speech.configure (also in native module interface and Jest mock).
2. Event payload id type changed from number to string.
3. speak and speakWithOptions now resolve with string (utterance ID) instead of void.
4. speakWithOptions is deprecated in favor of speak(text, options) (still present, but return type changed).

Fixes

1. Android queue handling now cleans up completed/error items and avoids stale indices.
2. Android stop/resume flow uses current utterance ID reliably.
3. iOS voice listing handles nil voices safely.
4. iOS error handling guards against missing utterances and cleans up metadata.
5. HighlightedText clamps ranges, tolerates unsorted highlights, handles empty inputs, and uses Text press handling reliably.

Performance

1. Android speech queue uses a LinkedHashMap to reduce queue scanning and improve lookup predictability.
2. HighlightedText avoids sorting when highlights are already ordered and reduces intermediate allocations.

Full Changelog: v1.3.2...v1.4.0