Explicit 202 response
The first request returns a durable job ID instead of holding an HTTP connection open during transcription.
Capslane turns captionless public YouTube videos into text through an asynchronous job, while keeping native caption requests available during heavier processing.
GET /v1/transcriptcurl 'https://capslane.com/v1/transcript?url=jNQXAC9IVRw&mode=generate' \
-H 'x-api-key: YOUR_API_KEY'Generated transcripts take longer than caption extraction. The API makes that difference explicit without forcing the client to maintain another response model.
The first request returns a durable job ID instead of holding an HTTP connection open during transcription.
Job status moves forward through download, processing and persistence without reversing the reported percentage.
Concurrent requests for the same transcript reuse one canonical generation job.
Caption extraction remains available while the worker processes a longer audio job.
Temporary media is removed after the transcript has been persisted.
Completed jobs return the same text, language and segment structure as native requests.
Request auto or generate mode and store the returned job ID when the API responds with 202.
Use the same API key on the job endpoint until the status becomes completed or failed.
Read the completed content immediately, then benefit from the cache on future requests.
Auto mode confirms that no usable native track exists, then queues local speech transcription and returns a job ID.
No. Start with GET /v1/transcript and poll GET /v1/transcript/:jobId when the first response has HTTP status 202.
Yes. Use mode=generate to use the audio path directly unless the canonical generated transcript is already cached.
No. Polling an accepted transcript job does not reserve another monthly request unit.
Start free with no credit card and move to a paid plan when the integration grows.