VoxaTrace

Voice AI today understands what you say: the words, the language, the text. But it's deaf to how you actually sound — pitch, timbre, emotion, vocal quality, rhythm, melody. Everything that makes a voice a voice, not just a transcript.

VoxaTrace is an on-device SDK that makes any application acoustically intelligent.

Eight years of R&D. Five million users in production. All running natively on Android and iOS, without a single server call.

What Speech AI Misses

Speech AI	VoxaTrace
"The user said 'hello'"	"The user sang A4 at 440 Hz with 92% confidence"
Words and language	Pitch, melody, rhythm, vocal quality
Transcription	Acoustic analysis
Cloud-dependent	On-device, real-time

What You Can Build

Application	What VoxaTrace Enables
Singing apps	Pitch detection, real-time scoring, performance feedback
Vocal training	Intonation analysis, progress tracking, guided exercises
Music education	Ear training, sight-singing evaluation, pitch matching
Voice games	Pitch as input — sing to jump, hum to control
Accessibility	Voice-based input beyond speech recognition
Health & wellness	Vocal health monitoring, breathing exercises

What You Get

┌────────────────────────────────────────────────────────────────────────┐
│                              VoxaTrace                                 │
├──────────────┬──────────────┬──────────────┬──────────────┬────────────┤
│    Sonix     │     Tona     │   Tessera    │   Accura     │  Calibra   │
│  Audio I/O   │    Pitch     │ Voice metrics│ Intonation   │  Singing   │
│              │              │              │ scoring      │  eval      │
├──────────────┼──────────────┼──────────────┼──────────────┼────────────┤
│ • Player     │ • Detection  │ • Breath     │ • EQ / JI    │ • LiveEval │
│ • Recorder   │ • Processing │ • Agility    │   per-note   │ • MelodyEval│
│ • Mixer      │ • Analysis   │ • Range      │   deviation  │ • NoteEval │
│ • Encoder    │   (histogram │ • Speaking   │ • 0–100      │ • VAD      │
│ • Decoder    │   transcr.)  │   pitch      │   scoring    │            │
│ • Metronome  │              │              │              │            │
│ • MIDI synth │              │              │              │            │
└──────────────┴──────────────┴──────────────┴──────────────┴────────────┘

           Common: MusicTheory (Hz/MIDI/cents conversions, shruti alignment)

Performance

Metric	Specification
Pitch detection latency	~64 ms (1024-sample window at 16 kHz, BALANCED)
Default frequency range	80 Hz – 1000 Hz (configurable per VoiceType)
SwiftF0 model range	46.875 Hz – 2093.75 Hz
Confidence threshold	0.0 – 1.0 (default 0.75 for BALANCED)
Sample rates	Auto-resampling to 16 kHz internally (ADR-017)
Minimum Android	API 26 (Android 8.0)
Minimum iOS	iOS 15

Hello, Pitch Detection

Kotlin

VT.initializeForServer("sk_live_…")  // see Authentication guide for mobile init

val detector = PitchDetection.createDetector()
val point = detector.detect(audioSamples, sampleRate = 16000)
println("${point.pitch} Hz @ ${(point.confidence * 100).toInt()}% confidence")
detector.close()

Swift

let detector = PitchDetection.createDetector()
let point = detector.detect(samples: audioSamples, sampleRate: 16000)
print("\(point.pitch) Hz @ \(Int(point.confidence * 100))% confidence")
detector.close()

Output:

440.0 Hz @ 92% confidence

Why VoxaTrace?

On-device, not cloud:

• Zero latency from network calls
• Works offline
• User data stays on device

Battle-tested:

• 8 years of R&D
• 5 million users in production
• Same API on Android and iOS

Three-tier API:

// Tier 1: Just works
val detector = PitchDetection.createDetector()

// Tier 2: Configurable
val detector = PitchDetection.createDetector(
    PitchDetectorConfig.Builder()
        .algorithm(PitchAlgorithm.SWIFT_F0)
        .build()
)

// Tier 3: Full control
val config = PitchDetectorConfig.PRECISE.copy(confidenceThreshold = 0.6f)

Read more about why VoxaTrace →

Start Building

5-Minute Quickstarts

• Android Quickstart – Pitch detector in Compose
• iOS Quickstart – Pitch detector in SwiftUI

Complete App Recipes

• Karaoke App – Play + record + score
• Tuner App – Chromatic tuner with cents display
• Voice Recorder – Record + encode + save
• Practice Tracker – Store scores, track progress

Understand the Concepts

• Pitch Detection – YIN vs SwiftF0, when to use each
• Live Evaluation – How singing scoring works
• Voice Activity – Detecting when someone is singing
• API Patterns – The three-tier API design

Module Deep Dives

• Sonix – Audio engine (player, recorder, mixer, encoder, decoder, …)
• Tona – Pitch detection / processing / analysis
• Tessera – Voice metrics (breath, agility, range, speaking pitch)
• Accura – Intonation analysis and 0–100 scoring
• Calibra – Singing evaluation (LiveEval / MelodyEval / NoteEval / VAD)
• Common: MusicTheory – Pitch ↔ MIDI ↔ note ↔ cents conversions, shruti alignment

Installation

// build.gradle.kts
dependencies {
    implementation("com.musicmuni:voxatrace:2.1.0")
}

// Package.swift
dependencies: [
    .package(url: "https://github.com/musicmuni/voxatrace", from: "2.1.0")
]

# Podfile
pod 'VoxaTrace', :podspec => 'https://raw.githubusercontent.com/musicmuni/voxatrace/main/VoxaTrace.podspec'

Full installation guide →

Support

• GitHub Issues – Bug reports and feature requests
• Demo Apps – Working examples