[roprgm]

Summary

• expose raw transcript deltas through onDelta
• move text composition into the demo with a functional state update
• avoid re-emitting the final transcript after streamed deltas
• document the native speech recognition limitation on iOS and iPadOS

Why

The provider previously accepted the full editor value and decided how to append transcription text. That coupled microphone/transcription behavior to textarea composition and made edits during recording vulnerable to stale value snapshots.

The component now emits deltas and leaves insertion, spacing, cursor handling, and alternate rendering strategies to the consumer. Internally it only tracks whether a delta arrived so a final-only transcriber can still emit its result once.

API

• replace value and onValueChange with optional onDelta
• remove the accumulated transcript from useChatVoiceInput()

Validation

• pnpm test — 31 tests passed
• pnpm typecheck — includes package and demo builds
• pnpm exec biome check src/provider.tsx demo/main.tsx tests/provider.test.tsx README.md
• git diff --check

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
chat-voice-input	Ready	Preview, Comment	Jul 25, 2026 9:57pm