VoiceStudio is the open-source, fully-local ElevenLabs alternative — voice cloning, voice design, video dubbing, dictation, transcription & audiobook creation in 646 languages.
Python aiaudiobookcudadubbingelevenlabs-alternativehuggingfacelocal-firstmlxomnivoice-studiospeech-to-texttext-to-speechtranscriptiontranslatettsvoice-aivoice-cloningvoice-generationvoicestudioworkflow
export const DEFAULT_OVERRIDES = {
numStep: null,
guidanceScale: null,
posTemp: null,
classTemp: null,
postprocess: null,
seed: null,
varyRepeats: false,
emoText: '',
emoAlpha: null,
// Seamless joins: null = the server default (zero gaps / trim off, preserving legacy renders).
lineGapMs: null,
paragraphGapMs: null,
trimEdges: null,
};
/**
* Lower the persisted overrides (+ language) into the snake_case request fields
* the backend expects. Only NON-default values are emitted, so an untouched
* panel adds nothing to the body → today's exact request. Shared by the full
* render and the per-chapter preview so both hit the same cache slot.
*/
export function overridesToRequest(overrides, language) {
const o = overrides || DEFAULT_OVERRIDES;
const body = {};
if (language && language !== 'Auto') body.language = language;
if (o.numStep != null) body.num_step = o.numStep;
if (o.guidanceScale != null) body.guidance_scale = o.guidanceScale;
if (o.posTemp != null) body.position_temperature = o.posTemp;
if (o.classTemp != null) body.class_temperature = o.classTemp;
if (o.postprocess != null) body.postprocess_output = o.postprocess;
if (o.seed != null) body.seed = o.seed;
if (o.varyRepeats) body.vary_repeats = true;
const emo = (o.emoText || '').trim();
if (emo) {
body.emo_text = emo;
if (o.emoAlpha != null) body.emo_alpha = o.emoAlpha;
}
if (o.lineGapMs != null) body.line_gap_ms = o.lineGapMs;
if (o.paragraphGapMs != null) body.paragraph_gap_ms = o.paragraphGapMs;
if (o.trimEdges != null) body.trim_edges = o.trimEdges;
return body;
}