Transcribe - Framelane

What this workflow does

Submits a POST /v1/tasks/transcribe job via AssemblyAI, polls until complete, and returns the transcript from the completed task’s result object. Word-level timestamps are always included so the output is ready to feed into word_animation in a render request (divide the millisecond timings by 1000 first — word_animation takes seconds).

Inputs

Name	Required	Description
`source_url`	required	URL of the audio or video file
`source_language`	optional	BCP-47 language code (e.g. `en`). Auto-detected if omitted.
`speaker_labels`	optional	Enable speaker diarisation. Default: `false`.

Output

Returns transcript — the completed task’s result object, including text, words[] (with text, start, end, confidence, speaker per word), and utterances[] when speaker labels are enabled. There is no file to download: output is null on a transcribe task and GET /v1/tasks/{id}/download answers 404.