SonixAudioUtils

Utility functions for manipulating audio data: concatenation, peak normalization, offline mixing, and frame/time conversions.

Quick Start

Kotlin

// Concatenate audio clips
val combined = SonixAudioUtils.concatenate(listOf(intro, verse, chorus))

// Peak-normalize audio
val normalized = SonixAudioUtils.normalize(decoded)

// Mix two tracks with custom gains
val mixed = SonixAudioUtils.mix(
    audioList = listOf(vocal, backing),
    gains = floatArrayOf(1.0f, 0.5f)
)

Swift

let combined = SonixAudioUtils.concatenate([intro, verse, chorus])
let normalized = SonixAudioUtils.normalize(audio: decoded)
let mixed = SonixAudioUtils.mix(audioList: [vocal, backing], gains: [1.0, 0.5])

Methods

concatenate

Join multiple audio clips end-to-end. All inputs must have the same sample rate and channel count.

val combined = SonixAudioUtils.concatenate(listOf(audio1, audio2, audio3))

normalize

Peak-normalize audio so the maximum absolute sample value reaches 1.0. Preserves dynamic range while maximizing signal level. Silent audio is returned unchanged.

val normalized = SonixAudioUtils.normalize(audio)

mix

Mix multiple mono audio tracks into a single output. Tracks are resampled to a common sample rate if needed, shorter tracks are zero-padded, and output is soft-clipped to prevent distortion.

For real-time playback mixing, use SonixMixer instead.

// Equal gain
val mixed = SonixAudioUtils.mix(listOf(vocal, backing))

// Custom gains and target sample rate
val mixed = SonixAudioUtils.mix(
    audioList = listOf(vocal, backing),
    gains = floatArrayOf(1.0f, 0.5f),
    targetSampleRate = 44100
)

Parameter	Type	Default	Description
audioList	List<AudioRawData>	required	Mono tracks to mix
gains	FloatArray?	null (1.0 for all)	Per-track gain factors
targetSampleRate	Int	First track's rate	Output sample rate in Hz

framesToTime / timeToFrames

Convert between frame indices and time in seconds.

val times = SonixAudioUtils.framesToTime(intArrayOf(0, 10, 20), sampleRate = 16000, hopLength = 512)
val frames = SonixAudioUtils.timeToFrames(floatArrayOf(0.0f, 0.32f), sampleRate = 16000, hopLength = 512)

framesToSegments

Convert a boolean frame mask to time segment pairs. Useful for extracting voiced regions from VAD output.

val mask = booleanArrayOf(false, true, true, true, false)
val segments = SonixAudioUtils.framesToSegments(mask, sampleRate = 16000, hopLength = 512)
// segments: [(0.032, 0.128)]

Method Summary

Method	Description
concatenate	Join audio clips sequentially
normalize	Peak-normalize to amplitude 1.0
mix	Offline mixing with gains and resampling
framesToTime	Frame indices to seconds
timeToFrames	Seconds to frame indices
framesToSegments	Boolean mask to time segments

Next Steps

• SonixToneGenerator — Generate synthetic waveforms
• SonixEncoder — Encode results to M4A, MP3, or WAV
• SonixDecoder — Decode audio files to AudioRawData
• SonixMixer — Real-time playback mixing