xsoulspace_inference_whisper_cpp_flutter 0.1.0-beta.1

whisper.cpp batch and low-latency incremental speech-to-text wrapper for xsoulspace_inference_core.

xsoulspace_inference_whisper_cpp_flutter

whisper.cpp wrapper for xsoulspace_inference_core, covering batch transcription plus low-latency incremental transcription sessions around whisper-stream.

Included API

• WhisperCppInferenceClient
• WhisperCppRealtimeSttSession
• WhisperCppModelPreset
• WhisperCppModelConfig
• WhisperCppRuntimeConfig
• WhisperCppRealtimeConfig
• WhisperCppAvailabilityProbe

Quickstart

final client = WhisperCppInferenceClient(
  modelConfig: const WhisperCppModelConfig(),
  runtimeConfig: const WhisperCppRuntimeConfig(
    librarySearchPaths: <String>['/opt/whisper'],
    modelsDirectory: '/opt/whisper-models',
  ),
);

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

Usage flow:

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

Testing

cd pkgs/xsoulspace_inference_whisper_cpp_flutter
flutter test