AssemblyAI

AssemblyAI transcribes audio/video. Transcripts are the core entity; paragraphs, sentences, subtitles are derived. Lemur is the AI summarization model.

Data Model

• Transcripts are completed transcriptions. List to get transcript id (tr_id). Paragraphs, sentences, subtitles are transcript-scoped.
• Lemur: AI summarization/Q&A. Pass transcript id as request_id — same value as tr_id.
• Hierarchy: Transcript → paragraphs/sentences/subtitle. All derived content needs transcript id.

Gotchas

• tr_id = request_id: For Lemur, transcript id doubles as request_id.
• Transcript status: List returns completed transcripts. Pending/processing may need polling.
• Lemur prompt: Custom prompt for summarization or Q&A.