getActiveStt static method

Future<SpeechRecognizer> getActiveStt({ PreferredBackend? preferredBackend, String? language, })

Get the active STT model as a ready-to-use SpeechRecognizer

Returns a SpeechRecognizer configured with runtime parameters. The model and tokenizer paths come from the active SttModelSpec.

Runtime parameters:

• preferredBackend: CPU or GPU preference (optional)
• language: the OUTPUT language for transcripts (Whisper only) — a bare lowercase code such as 'en' or 'de'. Sets SpeechRecognizer.language, and retargets the recognizer if one already exists, so it takes effect on every call and not just the first. Override a single transcription with SpeechRecognizer.transcribe's own language instead.

Throws:

• StateError if no active STT model is set
• ArgumentError for a malformed language, or for any language on a model whose decoder prompt has no language token (moonshine, parakeet)

Example:

// Install STT model first
await FlutterGemma.installStt()
  .modelFromNetwork('https://example.com/model.tflite')
  .tokenizerFromNetwork('https://example.com/tokenizer.json')
  .ofType(SttModelType.moonshine)
  .install();

// Create with default backend
final recognizer = await FlutterGemma.getActiveStt();

// Whisper: transcribe German, then French, on the SAME recognizer —
// nothing is reloaded between the two.
final de = await FlutterGemma.getActiveStt(language: 'de');
final german = await de.transcribe(germanPcm);
final french = await de.transcribe(frenchPcm, language: 'fr');

Implementation

static Future<SpeechRecognizer> getActiveStt({
  PreferredBackend? preferredBackend,
  String? language,
}) async {
  final manager = FlutterGemmaPlugin.instance.modelManager;
  final activeSpec = manager.activeSttModel;

  if (activeSpec == null) {
    throw StateError(
      'No active STT model set. Use FlutterGemma.installStt() first.',
    );
  }

  if (activeSpec is! SttModelSpec) {
    throw StateError(
      'Active model is not an SttModelSpec. '
      'Expected SttModelSpec, got ${activeSpec.runtimeType}',
    );
  }

  // Create SpeechRecognizer using active spec (paths resolved automatically)
  return await FlutterGemmaPlugin.instance.createSttModel(
    preferredBackend: preferredBackend,
    language: language,
  );
}