xsoulspace_inference_vosk_flutter 0.1.0-beta.1

Vosk small-model streaming and batch speech-to-text wrapper for xsoulspace_inference_core.

xsoulspace_inference_vosk_flutter

Vosk small-model wrapper for xsoulspace_inference_core, focused on low-footprint local realtime speech-to-text.

Included API

• VoskInferenceClient
• VoskRealtimeSttSession
• VoskAvailabilityProbe
• VoskModelConfig
• VoskRuntimeConfig
• VoskRealtimeConfig
• VoskModelPreset

Quickstart

1. Create provider config.
2. Create VoskInferenceClient or VoskRealtimeSttSession.
3. Run VoskAvailabilityProbe.probe().
4. Bind the injected session/client to generic UI from xsoulspace_inference_flutter.
5. Start transcribing.

final client = VoskInferenceClient(
  modelConfig: const VoskModelConfig(),
  runtimeConfig: const VoskRuntimeConfig(
    librarySearchPaths: <String>['/opt/vosk'],
    modelDirectory: '/opt/models',
  ),
);

final readiness = await VoskAvailabilityProbe(
  runtimeConfig: client.runtimeConfig,
  modelPathResolver: client.resolveModelPath,
).probe();

Testing

cd pkgs/xsoulspace_inference_vosk_flutter
flutter test