Skip to main content

Overview

The Audio Overview API generates conversational audio summaries from your completed notebook research. The endpoint creates a podcast-style discussion that synthesizes the research findings into an engaging audio format.

Generate Audio Overview

POST /research/audio-overview/{notebook_id}

Generate an audio overview from a notebook’s research summary

Request

notebook_id
string
required
The notebook ID containing completed research to generate audio from

Response

task_id
string
Unique identifier for the audio generation task
notebook_id
string
The notebook ID where results will be stored
status
string
Current task status: IN_QUEUE
message
string
Human-readable status message

Example Request

curl -X POST https://api.decipherit.com/research/audio-overview/nb_123456

Example Response

{
  "task_id": "task_audio_xyz789",
  "notebook_id": "nb_123456",
  "status": "IN_QUEUE",
  "message": "Audio overview task submitted and will be processed."
}

Error Responses

404 Not Found
object
500 Internal Server Error
object

Check Task Status

Once submitted, use the Get Task Status endpoint to monitor progress:
curl -X GET https://api.decipherit.com/research/task_audio_xyz789

Example Completed Task Response

{
  "task_id": "task_audio_xyz789",
  "notebook_id": "nb_123456",
  "topic": null,
  "sources": null,
  "status": "COMPLETED",
  "created_at": "2026-03-03T10:30:00Z",
  "result": {
    "transcript": [
      {
        "name": "Alex",
        "transcript": "Welcome to today's research overview. We'll be discussing quantum computing applications..."
      },
      {
        "name": "Jordan",
        "transcript": "Thanks Alex. The research reveals some fascinating insights into how quantum computers are revolutionizing..."
      }
    ]
  },
  "error": null,
  "completed_at": "2026-03-03T10:35:42Z",
  "failed_at": null
}

Audio Overview Structure

The generated audio overview contains a transcript with multiple speakers:
transcript
array
Array of conversation segments

How It Works

  1. Research Summary: The agent analyzes the notebook’s research summary
  2. Script Generation: Creates a conversational podcast-style script with multiple speakers
  3. Transcript Output: Returns the structured transcript for audio synthesis
  4. Storage: Results are stored in the notebook for later retrieval

Best Practices

  • Complete Research First: Ensure the notebook has completed research before generating audio
  • Async Processing: Audio generation is async - use task status polling to monitor progress
  • Error Handling: Implement retry logic for 500 errors
  • Notebook Validation: Always check that the notebook exists before submitting the task

Use Cases

  • Create podcast-style summaries of research topics
  • Generate audio briefings for on-the-go consumption
  • Produce engaging content from research findings
  • Accessibility: Convert written research into audio format

Build docs developers (and LLMs) love