xsoulspace_inference_flutter_tts

flutter_tts-backed text-to-speech implementation of xsoulspace_inference_core.

Supported task

  • InferenceTask.textToSpeech

Unsupported tasks (structuredText, speechToText) return task_unsupported.

Platform matrix (v1)

  • Android: artifact synthesis supported
  • iOS: artifact synthesis supported
  • macOS: returns task_unsupported for artifact-only TTS in v1
  • Linux/Windows/Web: not supported for this wrapper in v1

Install

dependencies:
  xsoulspace_inference_flutter_tts:
    path: ../xsoulspace_inference_flutter_tts
  xsoulspace_inference_core:
    path: ../xsoulspace_inference_core

Usage

import 'package:xsoulspace_inference_core/xsoulspace_inference_core.dart';
import 'package:xsoulspace_inference_flutter_tts/xsoulspace_inference_flutter_tts.dart';

final client = FlutterTtsInferenceClient();

final result = await client.infer(
  InferenceRequest.textToSpeech(
    text: 'Thanks for your message. Here is a corrected version.',
    workingDirectory: '/tmp',
    metadata: {
      'output_file_path': '/tmp/reply.wav',
      'output_mime_type': 'audio/wav',
    },
    voiceOptions: const InferenceVoiceOptions(
      locale: 'en-US',
      speechRate: 0.9,
      pitch: 1.0,
    ),
  ),
);

if (!result.success) {
  throw StateError('${result.error?.code}: ${result.error?.message}');
}

final artifact = result.data?.audioArtifact;

Notes

  • Wrapper is artifact-only in v1: no playback side effects and no streaming.
  • Set metadata['output_file_path'] to control where synthesized audio is written.
  • If no output path is provided, the wrapper creates tts_<timestamp>.wav/.caf in workingDirectory.

Testing

cd pkgs/xsoulspace_inference_flutter_tts
flutter test

Optional runtime-gated integration smoke:

cd pkgs/xsoulspace_inference_flutter_tts
FLUTTER_TTS_RUNTIME_INTEGRATION=1 flutter test test/integration/flutter_tts_runtime_integration_test.dart

Libraries

xsoulspace_inference_flutter_tts