CalibraVocalRange

Streaming vocal range analyzer for determining a singer's pitch range.

What is Vocal Range?

Vocal range is the span between a singer's lowest and highest notes. This analyzer detects the range using pedagogically-valid criteria based on ASHA 2018 guidelines and voice science research.

Use it for:

• Onboarding: Determine user's comfortable singing range

• Song selection: Match songs to user's range

• Progress tracking: See if range expands over time

• Key transposition: Calculate optimal key for user

When to Use

Quick Start

Kotlin

val analyzer = CalibraVocalRange.create()

// Feed audio from recorder
recorder.audioBuffers.collect { buffer ->
    analyzer.addAudio(buffer.toFloatArray())
}

// Get detected range
val range = analyzer.getRange()
if (range != null) {
    println("Range: ${range.lower.noteLabel} to ${range.upper.noteLabel}")
    println("Octaves: ${range.octaves}")
}

analyzer.release()

Swift

let analyzer = CalibraVocalRange.create()

// Feed audio from recorder
for await buffer in recorder.audioBuffersStream() {
    analyzer.addAudio(samples: buffer.toFloatArray())
}

// Get detected range
if let range = analyzer.getRange() {
    print("Range: \(range.lower.noteLabel) to \(range.upper.noteLabel)")
    print("Octaves: \(range.octaves)")
}

analyzer.release()

Guided Range Detection

For step-by-step range finding (e.g., "sing your lowest note"):

Kotlin

val analyzer = CalibraVocalRange.create()

// Phase 1: Detect lowest note
showPrompt("Sing your lowest comfortable note")
// ... user sings, feed audio via addAudio() ...
val lowestNote = analyzer.getStableNote()
analyzer.reset()  // Reset for next phase

// Phase 2: Detect highest note
showPrompt("Sing your highest comfortable note")
// ... user sings, feed audio via addAudio() ...
val highestNote = analyzer.getStableNote()

// Calculate range
val rangeOctaves = (highestNote.pitch.midiNote - lowestNote.pitch.midiNote) / 12f

analyzer.release()

Detection Criteria (ASHA 2018)

The analyzer uses these criteria for pedagogical validity:

• Minimum 1 second sustained pitch for note inclusion

• Confidence threshold of 0.5 minimum

• Stability: pitch deviation < 1 semitone within 50ms

• Percentile-based range extraction (5th-95th percentile)

Platform Notes

iOS

• Audio is typically 48kHz from microphone; internal resampling handles this

• Uses YIN algorithm for pitch detection (no external dependencies)

Android

• Audio varies by device (44.1kHz, 48kHz, 16kHz)

• Uses YIN algorithm for pitch detection (pure Kotlin)

Common Pitfalls

1. Forgetting to release: Call analyzer.release() to free resources

2. Not resetting between phases: Call reset() when starting new detection

3. Not enough sustained notes: User must hold notes for 1+ second

4. Background noise: High noise can trigger false detections

See also