Public release · 2026-10-01
• Introduces the public browser teleprompter beta with an editable script, manual pace and on-device English speech following. Scripts and audio stay in the browser; speech models download only after an explicit load action.
• Adds a reader-first layout with a single listening state, adjustable rendered-line lookahead, font and reading-zone controls, and horizontal or vertical mirroring.
• Adds deliberate backward realignment for sustained distinctive speech while preserving manual word and position seeking. Short batches can lag or miss words; larger-model performance and some real-device behavior remain under evaluation.
• Provides visible Credits, the adopted Community Use and Attribution License 1.0 for original code, separate third-party terms, and a version-specific editable-source download.