process method

@override Future<List<Map<String, dynamic>>> process(Float32List audioData, { double? step, })

Process audio data and return speaker segments

Returns a list of segments. For regular segmentation models:

{
  'speaker': int,    // Speaker index
  'start': double,   // Start time in seconds
  'stop': double,    // End time in seconds
}

Implementation

@override
Future<List<Map<String, dynamic>>> process(
  Float32List audioData, {
  double? step,
}) async {
  final snapshot = Float32List.fromList(audioData);
  await _pyannoteIsolateManager.start();
  return _pyannoteIsolateManager.sendInference(
    modelPath,
    snapshot,
    ortDylibPathOverride: fonnxOrtDylibPathOverride,
  );
}