Submit a text-to-speech batch
Queues up to 50 texts to be voiced, one clip per text, with one provider, model and voice. Returns as soon as the batch is accepted and its exact charge is held — poll GET /v1/tts/batch/{job_id} for the audio. Billed per 1,000 characters at your workspace’s text-to-speech rate; clips that fail to render are refunded.
Authentication
Bearer authentication of the form Bearer <token>, where token is your auth token.
Request
Texts to voice, one clip each, in order. Each is up to 5,000 characters; a longer one is rejected with 400 naming its index.
Provider that voices every clip. Resolved to your own workspace’s connection server-side; a provider your workspace has no connection to is rejected with 400.
Voice of the chosen provider, passed through unchanged: an ElevenLabs voice id (21m00Tcm4TlvDq8ikWAM), a Sarvam bulbul:v3 speaker (priya) or bulbul:v4-flash persona (ritu_hi_customer) matching tts_model, a Smallest lightning_v3.1_pro voice (mandar) or a Vakyam voice name (Archana). Not checked at submit — a wrong voice fails each clip with the provider’s message.
Must be selectable for tts_provider: eleven_multilingual_v2 (default), eleven_flash_v2_5, eleven_turbo_v2_5, eleven_v4 or eleven_v4_turbo for elevenlabs; bulbul:v3 (default) or bulbul:v4-flash for sarvam; lightning_v3.1_pro for smallest; raaga-v1 for vakyam. Omit it for the provider’s default.
Language name (Hindi, english) or locale code (hi-IN, en-IN), any case, normalized to a locale code. Omitted, Sarvam speaks en-IN, Smallest en and Vakyam hi-IN; ElevenLabs is not sent a language and speaks the language the text is written in. Text is spoken as written, not translated.
Response
Always pending on submit — rendering runs asynchronously.