CalibraSpeakingPitch

Natural speaking pitch detection for voice profiling.

What is Speaking Pitch?

Speaking pitch is the median fundamental frequency of a person's voice when speaking naturally. It represents their "home base" vocal frequency.

Use it for:

• Voice profiling: Establish user's natural pitch range

• Shruti suggestion: Recommend musical tonic based on voice

• Voice type classification: Soprano, tenor, bass, etc.

• Voice health tracking: Monitor changes over time

Note: Speaking pitch is different from:

• Singing range: Full high-to-low capability (use CalibraVocalRange)

• Shruti/tonic: Musical reference note (calculated from range)

When to Use

Quick Start

Kotlin

// From audio samples (16kHz mono)
val speakingPitch = CalibraSpeakingPitch.detectFromAudio(audioSamples)

if (speakingPitch > 0) {
    println("Speaking pitch: $speakingPitch Hz")
    val note = CalibraMusic.hzToNoteLabel(speakingPitch)
    println("Closest note: $note")
}

// Or from existing pitch contour
val contour = pitchExtractor.extract(audio, 16000)
val speakingPitch = CalibraSpeakingPitch.detectFromPitch(contour.toPitchesArray())

Swift

// From audio samples (16kHz mono)
let speakingPitch = CalibraSpeakingPitch.detectFromAudio(audioMono: audioSamples)

if speakingPitch > 0 {
    print("Speaking pitch: \(speakingPitch) Hz")
    let note = CalibraMusic.hzToNoteLabel(speakingPitch)
    print("Closest note: \(note)")
}

// Or from existing pitch contour
let contour = pitchExtractor.extract(audio: audio, sampleRate: 16000)
let speakingPitch = CalibraSpeakingPitch.detectFromPitch(pitchesHz: contour.toPitchesArray())

Typical Speaking Pitches

Platform Notes

iOS/Android

• Accepts any sample rate; internally resamples to 16kHz if needed (ADR-017)

• Uses median-based detection for robustness against outliers

• Returns -1 if detection fails (not enough voiced audio)

Common Pitfalls

1. Singing instead of speaking: This detects speaking pitch, not singing

2. Not enough audio: Need several seconds of natural speech

3. Background noise: High noise levels affect detection

See also