Create Voice
Beta. Creates a cloned voice from a reference audio clip in one shot. The whole pipeline runs inline: the audio is stored, validated for quality, transcribed via ASR, and cloned - the response returns the new voice_id and the auto-generated transcript (reference_text) together. No polling or status endpoint is needed; the voice is ready to use immediately with POST /voices/clone.
A background job improves the reference audio (noise reduction) after the voice is created; this does not block the response.
Accepts multipart/form-data - send the audio as the file upload and everything else as form fields.
Base URL: https://api.sarvam.ai.
Auth: send your key in the api-subscription-key header.
Limits: max file size 50 MB. Voice count is capped by your subscription tier (e.g. 3 voices on the starter tier) - the API returns 403 TIER_LIMIT_REACHED when the cap is hit.
Billing: charged against the voice cloning API. See the pricing page.
Authentication
Request
Reference audio file used to create the cloned voice. Supported formats: WAV, MP3, FLAC, OPUS, AAC, MULAW, ALAW, LINEAR16. WAV is recommended.
Limits: max file size 50 MB. Recommended duration is 10-15 seconds, single speaker, minimal background noise.
Display name for the cloned voice (1-100 characters).
BCP-47 language code of the reference audio (e.g. en-IN, hi-IN). The transcript is auto-generated via ASR; you do not need to pass it.
Response
Voice created successfully. Usable immediately with POST /voices/clone.
success