VoxaTrace/com.musicmuni.voxatrace.tona.model/ContourExtractorConfig

ContourExtractorConfig

data class ContourExtractorConfig(val preset: PitchPreset = PitchPreset.BALANCED, val algorithm: PitchAlgorithm = PitchAlgorithm.SWIFT_F0, val sampleRate: Int = 16000, val hopMs: Int = 10, val cleanup: PitchProcessingConfig = PitchProcessingConfig.SCORING, val voiceType: VoiceType = VoiceType.Auto, val quietHandling: QuietHandling = QuietHandling.NORMAL, val strictness: DetectionStrictness = DetectionStrictness.BALANCED)

Configuration for batch pitch contour extraction (ADR-001).

Passed to PitchDetection.createContourExtractor. Controls which algorithm is used, the hop size between frames, and the post-processing cleanup applied.

Usage Tiers

Tier 1: Presets (80% of users)

val extractor = PitchDetection.createContourExtractor(
    ContourExtractorConfig.SCORING,
    modelProvider = { ModelLoader.loadSwiftF0() }
)

Tier 2: Builder (15% of users)

val config = ContourExtractorConfig.Builder()
    .preset(ContourExtractorConfig.SCORING)
    .sampleRate(16000)
    .hopMs(10)
    .build()
val extractor = PitchDetection.createContourExtractor(config, modelProvider = { ... })

Tier 3: .copy() (5% of users)

val config = ContourExtractorConfig.SCORING.copy(hopMs = 5)

Constructors

ContourExtractorConfig

constructor(preset: PitchPreset = PitchPreset.BALANCED, algorithm: PitchAlgorithm = PitchAlgorithm.SWIFT_F0, sampleRate: Int = 16000, hopMs: Int = 10, cleanup: PitchProcessingConfig = PitchProcessingConfig.SCORING, voiceType: VoiceType = VoiceType.Auto, quietHandling: QuietHandling = QuietHandling.NORMAL, strictness: DetectionStrictness = DetectionStrictness.BALANCED)