---
title: "Audio to Insights"
description: "Run the complete asynchronous audio processing workflow."
---

This example uploads an audio file, waits for transcription, and generates structured Insights.

```python
import os
import time
import uuid

import requests

BASE_URL = "https://audream-api.tulingbc.com"
API_KEY = os.environ["AUDREAM_API_KEY"]
HEADERS = {"Authorization": f"Bearer {API_KEY}"}

note_id = str(uuid.uuid4())
audio_path = "meeting.m4a"

with open(audio_path, "rb") as audio:
    response = requests.post(
        f"{BASE_URL}/v1/notes/{note_id}/transcription",
        headers=HEADERS,
        files={"file": (audio_path, audio)},
        data={
            "title": "Product review",
            "language": "auto",
            "auto_speaker_labeling_enabled": "true",
        },
        timeout=120,
    )
response.raise_for_status()

while True:
    response = requests.get(
        f"{BASE_URL}/v1/notes/{note_id}/transcription",
        headers=HEADERS,
        timeout=30,
    )
    if response.status_code == 200:
        transcription = response.json()
        break
    if response.status_code == 202:
        time.sleep(2)
        continue
    response.raise_for_status()

response = requests.post(
    f"{BASE_URL}/v1/notes/{note_id}/insights",
    headers=HEADERS,
    json={
        "result": transcription,
        "title": "Product review",
    },
    timeout=900,
)

if response.status_code == 202:
    while True:
        response = requests.get(
            f"{BASE_URL}/v1/notes/{note_id}/insights",
            headers=HEADERS,
            timeout=30,
        )
        if response.status_code == 200:
            break
        if response.status_code != 202:
            response.raise_for_status()
        time.sleep(2)

response.raise_for_status()
insights = response.json()
print(insights.get("abstract"))
```

## Supported audio formats

The transcription endpoint accepts files ending in:

- `.wav`
- `.opus`
- `.ogg`
- `.m4a`
- `.mp3`
- `.aac`

## Optional artifacts

After transcription completes, you can generate an additional artifact without replacing the note's Insights:

```python
response = requests.post(
    f"{BASE_URL}/v1/notes/{note_id}/artifacts/deep_research",
    headers=HEADERS,
    json={"language": "auto", "regenerate": False},
    timeout=30,
)
```

Supported artifact kinds are `epiphany`, `deep_research`, and `podcast`.