Create Dubbing Job | Sarvam API Docs

Creative Agents - Dubbing

Create a dubbing job that translates a source video or audio file into one or more Indian languages, with optional voice cloning, watermark control, and translation-style (register) control.

Base URL: https://api.sarvam.ai/dubbing. Auth: send your key in the api-subscription-key header, not Authorization: Bearer.

This call only creates the job. It does not accept the media file. The response returns data.job_id and a short-lived signed data.upload_url. After that:

PUT the raw media bytes to upload_url with headers Content-Type: <mime> (e.g. video/mp4) and x-ms-blob-type: BlockBlob.
POST /jobs/{job_id}/start to begin the pipeline.

A single job dubs into every language in target_langs, so you don’t need a separate job per language.

Authentication

api-subscription-keystring

API Key authentication via header

Request

This endpoint expects an object.

src_langenumRequired

Source language of the input media (BCP-47), or auto to detect it. Pass the language when you know it: detection is not always accurate. See Specify Language Codes.

target_langslist of enumsRequired

One or more target languages to dub into. A single job dubs into all of them.

export_optionslist of enumsOptional

Which formats to auto-produce per target language. When provided, only the listed formats are auto-exported. Accepted values: video (dubbed MP4), audio (dubbed voice over the background, .wav), audio_vocals (dubbed voice alone, .wav), srt and vtt (subtitles of the dubbed text). When omitted, only the dubbed video (MP4) is produced; request audio and subtitles explicitly. Audio-only sources cannot produce a video, so video resolves to audio for them.

Allowed values:

voice_cloningbooleanOptionalDefaults to true

Clone the original speaker’s voice. Set false to use a preset voice_id instead.

voice_idstring or nullOptional

Preset (prebuilt) voice to use when voice_cloning is false. Single speaker only.

pace_presetenum or nullOptional

Speech pace preset, used when voice_cloning is false.

Allowed values:

num_speakersinteger or nullOptionalDefaults to 1

Number of speakers in the source audio.

disable_watermarkbooleanOptionalDefaults to false

Set true for a watermark-free export.

registerenum or nullOptional

Translation-style / tone register.

editor_flowbooleanOptionalDefaults to false

Keep false for API integrations so exports fire automatically once TTS finishes. When true, auto-export is suppressed: the pipeline completes but export-status stays empty until exports are triggered manually in Creator Studio. Billing: true charges the editor-flow rate (₹80/min on Starter — double the default API rate). See Pricing for plan-specific rates.

job_namestring or nullOptional

Human-readable label to identify the job later. Recommended (e.g. the source file’s name).

Response

Successful Response

statusstring

Response status, e.g. success.

messagestring

dataobject

Errors

400

Bad Request Error

401

Unauthorized Error

500

Internal Server Error