TesseraAgility

Public facade for vocal agility analysis.

What is Vocal Agility?

Agility measures how dynamically a singer varies pitch — fast ornaments, gamakas, runs. The 10-stage pipeline computes log-derivative RMS of pitch (captures speed of pitch change) and oscillation RMS (captures regularity), then a linear classifier maps these to a 0–1 agility score.

When to Use

Quick Start

Kotlin

val score = TesseraAgility.computeScore(contour)
println("Agility: ${score.scores.first()}")

// Per-segment scoring
val segments = listOf(0f to 5f, 5f to 10f)
val segScores = TesseraAgility.computeScore(contour, segments = segments)

Swift

let score = TesseraAgility.computeScore(contour: contour)
print("Agility: \(score.scores.first ?? 0)")

Common Pitfalls

1. Contour must have ≥ 2 samples: Throws per ADR-022.

2. Scores below 0.5 indicate little agility: The sigmoid output ranges 0–1; a monotone contour scores near 0. Meaningful agility pushes scores above ~0.5.

3. Segment times are in seconds: Pair(startSec, endSec). Segments with no data points in range score near 0 (not 0.5). Only zero/negative- duration segments return an explicit 0.5 fallback.

See also