voice_flutter

Graph-backed Flutter voice input/output bridges and optional widgets driven by voice_core state.

GraphVoiceInput routes one capture to independent streaming STT and VAD sessions. GraphVoiceSpeechOutput routes every synthesized source to playback and optional recording, metering, or analysis branches.

import 'package:voice_flutter/voice_flutter.dart';

final input = GraphVoiceInput(
  source: microphoneSource,
  streamingSpeechToText: selectedStt,
  voiceActivityDetection: selectedVad,
  recognitionOptions: recognitionOptions,
);

When half-duplex output gates recognition, only the STT session is replaced; capture and VAD remain active for barge-in. Provider selection stays in the application composition layer.