Create Chapter Summaries Using LLM Gateway

1
import requests
2
import time
3
4
base_url = "https://api.assemblyai.com"
5
headers = {"authorization": "<YOUR_API_KEY>"}
6
7
# Step 1: Transcribe and get paragraphs
8
with open("./my-audio.mp3", "rb") as f:
9
    response = requests.post(base_url + "/v2/upload", headers=headers, data=f)
10
11
upload_url = response.json()["upload_url"]
12
data = {"audio_url": upload_url}
13
14
response = requests.post(base_url + "/v2/transcript", json=data, headers=headers)
15
transcript_id = response.json()['id']
16
polling_endpoint = base_url + "/v2/transcript/" + transcript_id
17
18
while True:
19
    transcription_result = requests.get(polling_endpoint, headers=headers).json()
20
    if transcription_result['status'] == 'completed':
21
        break
22
    elif transcription_result['status'] == 'error':
23
        raise RuntimeError(f"Transcription failed: {transcription_result['error']}")
24
    else:
25
        time.sleep(3)
26
27
# Get paragraphs
28
paragraphs = requests.get(polling_endpoint + '/paragraphs', headers=headers).json()['paragraphs']
29
30
# Step 2: Combine paragraphs and create chapter summaries
31
combined_paragraphs = []
32
step = 2  # Adjust as needed if you want combined paragraphs to be shorter or longer in length.
33
34
# Combine paragraphs into groups, finding the appropriate timestamps and combining all their text into one string.
35
for i in range(0, len(paragraphs), step):
36
    paragraph_group = paragraphs[i : i + step]
37
    start = paragraph_group[0]['start']
38
    end = paragraph_group[-1]['end']
39
    text = ""
40
    for paragraph in paragraph_group:
41
        text += f"{paragraph['text']} "
42
    combined_paragraphs.append(f"Paragraph: {text} Start: {start} End: {end}")
43
44
results = []
45
46
# Step 3: Generate summaries with LLM Gateway
47
for paragraph in combined_paragraphs:
48
    prompt = "Summarize this text as a whole and provide start and end timestamps."
49
50
    llm_gateway_data = {
51
        "model": "claude-sonnet-4-5-20250929",
52
        "messages": [
53
            {"role": "user", "content": f"{prompt}\n\n{paragraph}"}
54
        ],
55
        "max_tokens": 1000
56
    }
57
58
    response = requests.post(
59
        "https://llm-gateway.assemblyai.com/v1/chat/completions",
60
        headers=headers,
61
        json=llm_gateway_data
62
    )
63
64
    result = response.json()["choices"][0]["message"]["content"]
65
    results.append(result)
66
67
for result in results:
68
    print(f"{result}\n")

1
import requests
2
import time
3
4
base_url = "https://api.assemblyai.com"
5
headers = {"authorization": "<YOUR_API_KEY>"}
6
7
with open("./my-audio.mp3", "rb") as f:
8
    response = requests.post(base_url + "/v2/upload", headers=headers, data=f)
9
10
upload_url = response.json()["upload_url"]
11
data = {"audio_url": upload_url}  # You can also use a URL to an audio or video file on the web
12
13
response = requests.post(base_url + "/v2/transcript", json=data, headers=headers)
14
transcript_id = response.json()['id']
15
polling_endpoint = base_url + "/v2/transcript/" + transcript_id
16
17
while True:
18
    transcription_result = requests.get(polling_endpoint, headers=headers).json()
19
    if transcription_result['status'] == 'completed':
20
        print(f"Transcript ID:", transcript_id)
21
        break
22
    elif transcription_result['status'] == 'error':
23
        raise RuntimeError(f"Transcription failed: {transcription_result['error']}")
24
    else:
25
        time.sleep(3)

1
paragraphs = requests.get(polling_endpoint + '/paragraphs', headers=headers).json()['paragraphs']
2
combined_paragraphs = []
3
step = 2  # Adjust as needed if you want combined paragraphs to be shorter or longer in length.
4
5
# Combine paragraphs into groups, finding the appropriate timestamps and combining all their text into one string.
6
for i in range(0, len(paragraphs), step):
7
    paragraph_group = paragraphs[i : i + step]
8
    start = paragraph_group[0]['start']
9
    end = paragraph_group[-1]['end']
10
    text = ""
11
    for paragraph in paragraph_group:
12
        text += f"{paragraph['text']} "
13
    combined_paragraphs.append(f"Paragraph: {text} Start: {start} End: {end}")

1
results = []
2
3
for paragraph in combined_paragraphs:
4
    prompt = "Summarize this text as a whole and provide start and end timestamps."
5
6
    llm_gateway_data = {
7
        "model": "claude-sonnet-4-5-20250929",
8
        "messages": [
9
            {"role": "user", "content": f"{prompt}\n\n{paragraph}"}
10
        ],
11
        "max_tokens": 1000
12
    }
13
14
    response = requests.post(
15
        "https://llm-gateway.assemblyai.com/v1/chat/completions",
16
        headers=headers,
17
        json=llm_gateway_data
18
    )
19
20
    result = response.json()["choices"][0]["message"]["content"]
21
    results.append(result)
22
23
for result in results:
24
    print(f"{result}\n")