API reference

Construction

KittenTTS.create({
  KittenTTSConfig? config,
  ProgressHandler? onProgress,
  AudioPlayer? player,
  bool forceRedownload = false,
})

Instance methods

Method	Purpose
generate(text, {voice, speed})	Return one KittenTTSResult
stream(text, {voice, speed})	Yield sentence-sized results
speak(text, {voice, speed, options})	Generate and play
play(result, [options])	Play existing audio
createPlaybackQueue()	Create FIFO queue
stopSpeaking() / stop()	Stop playback
pauseSpeaking()	Pause configured player
resumeSpeaking()	Resume configured player
isSpeaking	Current player state
dispose()	Release engine and phonemizer resources

Result

samples, sampleRate, duration, voice, effectiveSpeed, inputText, wordTimings, wavData(), and wavBase64() are supported. MP3 helpers currently throw.

Cache methods

cacheInfo, predownload, validateAssets, redownloadModel, and clearModelCache are static.

Errors

KittenTTSError exposes a code and message. Codes include emptyInput, downloadFailed, invalidModelData, phonemizerFailed, inferenceFailed, and playbackFailed.

The package's exported Dart types are the authoritative source for exact signatures.