//clika-runtime/io.clika.modelverse/SpeakOptions

SpeakOptions

[common]
data class SpeakOptions(val exaggeration: Float? = null, val temperature: Float? = null, val cfgScale: Float? = null, val steps: Int = 0, val seed: Long = -1, val maxSpeechTokens: Int = 0, val language: String? = null)

The synthesis knobs of one request. A null knob, or the value SpeakKnob.keepsDefault names for it, keeps the family's own trained default (the chatterbox family: exaggeration 0.5, temperature 0.6, guidance 1.0, ten refinement steps, no seed, the text-length budget). A value under a knob's SpeakKnob.min is refused before the model runs.

Constructors

SpeakOptions	[common] constructor(exaggeration: Float? = null, temperature: Float? = null, cfgScale: Float? = null, steps: Int = 0, seed: Long = -1, maxSpeechTokens: Int = 0, language: String? = null)

Properties

Name	Summary
cfgScale	[common] val cfgScale: Float? = null The classifier-free-guidance strength, 0 or more; 0 turns the guidance off.
exaggeration	[common] val exaggeration: Float? = null Emotion exaggeration, 0 or more: about 0.25 reads neutral, 1.0 intense.
language	[common] val language: String? = null The utterance's language, by the family's own code (en, fr); null keeps the model's default. A model that interprets no language refuses a set one, and the request ends in SpeakListener.onError.
maxSpeechTokens	[common] val maxSpeechTokens: Int = 0 The speech-token budget ceiling, 0 or more; 0 keeps the plan the text's length sets. A smaller value may cut a long utterance, which the model refuses rather than truncates.
seed	[common] val seed: Long The random seed; 0 or more makes the synthesis repeatable, -1 seeds nothing.
steps	[common] val steps: Int = 0 Refinement steps per decode block, 0 or more; 0 keeps the family's default.
temperature	[common] val temperature: Float? = null The sampling temperature of the speech-token decode, 0 or more; 0 decodes greedily.