Skip to main content

Base URL

Authentication

All API requests require an API key passed in the X-API-Key header.

Endpoints

Upload Endpoints

Upload Media

Upload by file, URL, or request a signed upload URL

Notify Upload Complete

Start verification after a signed upload finishes

Get Upload Status

Poll signed-upload verification status

Text Generation Endpoints

Generate Text

Generate text using a media_id, URL, or direct file upload

Audio Generation Endpoints

Generate Audio

Generate an audio track using a media_id, URL, or direct file upload

Video Generation Endpoints

Generate Video

Generate a video using a media_id, URL, or direct file upload

Results Endpoint

Get Results

Poll for job results and download outputs

Quick Example

1

Upload

2

Generate

3

Get Results

Languages and Voices

For the full list of 53 supported language codes and 31 voice IDs, including which voices cover which languages, see Languages and Voices.

Video Categories

Auto • Educational Lecture • Educational Kids • Government Meeting • Documentary • Narrative Story • Social Media • Tutorial/How-To • Vlog • Commercial/Advertisement • News • Entertainment • Home Video • Video Call

Output Formats

Every generation endpoint returns the text descriptions in the response. Set format (default vtt) to choose how that text is serialized: The audio endpoint additionally returns a WAV track and the video endpoint additionally returns an MP4. For text-only output, see the text endpoint.

Error Codes

Common Error Payloads

Timeouts and Retries

Every generate request we accept starts its own job, and that job is charged. A request rejected with one of the errors above creates no job and costs nothing, and so does a job that fails its upload checks. If your client gives up waiting and sends the same request again, you get, and pay for, a second job.
  • Generate requests return a job_id within a few seconds, including requests that pass a video URL. Use a client timeout of at least 60 seconds.
  • A URL request that also sends your own description_content downloads the file before it responds. For videos over an hour, allow at least 5 minutes.
  • upload_media with a URL downloads the file before it responds. For videos over an hour, allow at least 5 minutes.
  • Don’t resend a generate request automatically after a timeout. Poll the jobs you already have with /get_results instead.
  • Retrying is safe after a 429, after 409 upload_not_ready, and for upload_media, because no job was created and uploads are not charged.

Job Failure Errors

When a job that was successfully accepted later fails during processing, polling the job (via /get_results) returns status: "failed" along with a stable error code and a human-readable message. Switch on the error code to handle failures programmatically; show message in your UI.
These codes are returned in the error field of /get_results for failed jobs and are stable across releases. Additional codes may be added in future; treat unknown codes as internal_error.

Rate Limits

Limits are enforced per team across all API keys. Enterprise plans can request custom allocations; contact support if you need higher throughput.

Rate Limit Exceeded Response

When you exceed the rate limit, you’ll receive a 429 Too Many Requests response:
The retry_after field indicates how many seconds to wait before retrying.

Need Help?

Support

Email our team

Dashboard

View usage