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
// Minimal helpers for reading a POST Server-Sent-Events stream via
// fetch() + response.body.getReader() (EventSource is GET-only, so it can't
// be used for our POST /audiobook stream). Pure + unit-tested so the
// buffer/line handling — the easy thing to get subtly wrong — is verified.
/**
* Split an accumulated decoded-text buffer into complete lines plus the
* trailing (possibly incomplete) remainder. Feed the remainder back in with
* the next chunk.
*
* @param {string} buffer
* @returns {{ lines: string[], rest: string }}
*/
export function splitSSEBuffer(buffer) {
const parts = (buffer || '').split('\n');
const rest = parts.pop() ?? ''; // last part has no trailing newline yet
return { lines: parts, rest };
}
/**
* Parse one SSE line into its JSON event object, or null when the line isn't
* a (well-formed) ``data:`` line. Tolerates ``data:`` with or without the
* conventional trailing space, blank lines, and malformed JSON (returns null
* rather than throwing, so one bad frame can't kill the read loop).
*
* @param {string} line
* @returns {object | null}
*/
export function parseSSELine(line) {
if (typeof line !== 'string' || !line.startsWith('data:')) return null;
const payload = line.slice(5).trim(); // after 'data:'
if (!payload) return null;
try {
return JSON.parse(payload);
} catch {
return null;
}
}