Diarize

What you can do

Connect this agent connector to let your agent:

• Get get — Retrieve the current status of a transcription job by its job ID
• Transcript download — Download the transcript output for a completed transcription job in JSON, TXT, SRT, or VTT format, including speaker diarization, segments, and word-level timestamps
• Create create — Submit a new transcription and diarization job for an audio or video URL (YouTube, X, Instagram, TikTok)

Authentication

This connector uses Bearer Token authentication. Scalekit securely stores the token and injects it into API requests on behalf of your users. Your agent code never handles tokens directly — you only pass a connectionName and a user identifier.

Set up the connector

Register your Diarize API key with Scalekit so it can authenticate and proxy transcription requests on behalf of your users. Unlike OAuth connectors, Diarize uses API key authentication — there is no redirect URI or OAuth flow.

1. Get a Diarize API key

   • Sign in to diarize.io and go to Settings → API Keys.

   • Click + Create New Key, give it a name (e.g., Agent Auth), and confirm.

   • Copy the key value — store it securely, as you will not be able to view it again.

   

2. Create a connection in Scalekit

   • In Scalekit dashboard, go to Agent Auth → Create Connection. Find Diarize and click Create.

   • Note the Connection name — you will use this as connection_name in your code (e.g., diarize).

   • Click Save.

   

3. Add a connected account

   Connected accounts link a specific user identifier in your system to a Diarize API key. Add accounts via the dashboard for testing, or via the Scalekit API in production.

   Via dashboard (for testing)

   • Open the connection you created and click the Connected Accounts tab → Add account.

   • Fill in:

     • Your User’s ID — a unique identifier for this user in your system (e.g., user_123)
     • API Key — the Diarize API key you copied in step 1

   • Click Create Account.

   

   Via API (for production)

   Node.js

   // Never hard-code API keys — read from secure storage or user input
   const diarizeApiKey = getUserDiarizeKey(); // retrieve from your secure store
   
   await scalekit.actions.upsertConnectedAccount({
     connectionName: 'diarize',
     identifier: 'user_123',       // your user's unique ID
     credentials: { token: diarizeApiKey },
   });

   Python

   # Never hard-code API keys — read from secure storage or user input
   diarize_api_key = get_user_diarize_key()  # retrieve from your secure store
   
   scalekit_client.actions.upsert_connected_account(
       connection_name="diarize",
       identifier="user_123",
       credentials={"token": diarize_api_key}
   )

   Production usage tip

   In production, call upsert_connected_account (Python) / upsertConnectedAccount (Node.js) when a user connects their Diarize account — for example, after they paste their API key into a settings page in your app.

Supported media sources

Diarize supports YouTube, X (Twitter), Instagram, and TikTok URLs. Direct audio or video file URLs are not supported — the URL must point to a public post on one of these platforms.

Code examples

Connect a user’s Diarize account and transcribe audio and video content through Scalekit tools. Scalekit handles API key storage and tool execution automatically — you never handle credentials in your application code.

Diarize is primarily used through Scalekit tools. Use execute_tool to submit transcription jobs, poll for completion, and download results in any supported format.

Tool calling

Use this connector when you want an agent to transcribe and diarize audio or video from YouTube, X, Instagram, or TikTok.

• Use diarize_create_transcription_job to submit a URL for transcription. Returns an id (job ID) and an estimatedTime (in seconds) for how long processing will take.
• Use diarize_get_job_status to poll until status is COMPLETED or FAILED. Use estimatedTime to set a sensible timeout — do not give up before that time has elapsed.
• Use diarize_download_transcript to retrieve the result once complete. Choose json for structured speaker diarization data, or txt, srt, vtt for plain-text and subtitle formats.

Python

examples/diarize_transcribe.py

import os, time
from scalekit.client import ScalekitClient

scalekit_client = ScalekitClient(
    client_id=os.environ["SCALEKIT_CLIENT_ID"],
    client_secret=os.environ["SCALEKIT_CLIENT_SECRET"],
    env_url=os.environ["SCALEKIT_ENV_URL"],
)

connected_account = scalekit_client.actions.get_or_create_connected_account(
    connection_name="diarize",
    identifier="user_123",
).connected_account

# Step 1: Submit a transcription job
create_result = scalekit_client.actions.execute_tool(
    tool_name="diarize_create_transcription_job",
    connected_account_id=connected_account.id,
    tool_input={
        "url": "https://www.youtube.com/watch?v=example",
        "language": "en",   # optional — omit for auto-detection
        "num_speakers": 2,  # optional — improves speaker diarization
    },
)
job_id = create_result.result["id"]
estimated_seconds = create_result.result.get("estimatedTime", 120)
deadline = time.time() + estimated_seconds * 2
print(f"Job {job_id} submitted. Estimated: {estimated_seconds}s")

# Step 2: Poll until complete
while True:
    if time.time() > deadline:
        raise TimeoutError(f"Job {job_id} timed out after {estimated_seconds * 2}s")
    time.sleep(15)
    status_result = scalekit_client.actions.execute_tool(
        tool_name="diarize_get_job_status",
        connected_account_id=connected_account.id,
        tool_input={"job_id": job_id},
    )
    status = status_result.result["status"]
    print("Status:", status)
    if status == "COMPLETED":
        break
    if status == "FAILED":
        raise RuntimeError(f"Job {job_id} failed")

# Step 3: Download the diarized transcript
transcript_result = scalekit_client.actions.execute_tool(
    tool_name="diarize_download_transcript",
    connected_account_id=connected_account.id,
    tool_input={"job_id": job_id, "format": "json"},
)
# handle the transcript_result

Node.js

examples/diarize_transcribe.ts

import { ScalekitClient } from '@scalekit-sdk/node';
import 'dotenv/config';

const scalekit = new ScalekitClient(
  process.env.SCALEKIT_ENV_URL!,
  process.env.SCALEKIT_CLIENT_ID!,
  process.env.SCALEKIT_CLIENT_SECRET!
);
const actions = scalekit.actions;

const { connectedAccount } = await actions.getOrCreateConnectedAccount({
  connectionName: 'diarize',
  identifier: 'user_123',
});

// Step 1: Submit a transcription job
const createResult = await actions.executeTool({
  toolName: 'diarize_create_transcription_job',
  connectedAccountId: connectedAccount.id,
  toolInput: {
    url: 'https://www.youtube.com/watch?v=example',
    language: 'en',   // optional — omit for auto-detection
    num_speakers: 2,  // optional — improves speaker diarization
  },
});
const jobId = createResult.data.id;
const estimatedSeconds = createResult.data.estimatedTime ?? 120;
const deadline = Date.now() + estimatedSeconds * 2 * 1000;
console.log(`Job ${jobId} submitted. Estimated: ${estimatedSeconds}s`);

// Step 2: Poll until complete
let status = 'PENDING';
while (status !== 'COMPLETED' && status !== 'FAILED') {
  if (Date.now() > deadline) throw new Error(`Job ${jobId} timed out after ${estimatedSeconds * 2}s`);
  await new Promise(r => setTimeout(r, 15_000));
  const statusResult = await actions.executeTool({
    toolName: 'diarize_get_job_status',
    connectedAccountId: connectedAccount.id,
    toolInput: { job_id: jobId },
  });
  status = statusResult.data.status;
  console.log('Status:', status);
}
if (status === 'FAILED') throw new Error(`Job ${jobId} failed`);

// Step 3: Download the diarized transcript
const transcriptResult = await actions.executeTool({
  toolName: 'diarize_download_transcript',
  connectedAccountId: connectedAccount.id,
  toolInput: { job_id: jobId, format: 'json' },
});
// handle the transcriptResult

Polling guidance

The estimatedTime field (in seconds) tells you how long processing is expected to take. For a 49-minute episode, estimatedTime may be around 891 seconds (~15 minutes). Wait at least that long before treating the job as timed out.

Scalekit tools

Tool list

diarize_create_transcription_job Submit a new transcription and diarization job for an audio or video URL (YouTube, X, Instagram, TikTok). Returns a job ID that can be used to check status and download results. 5 params ▾

Submit a new transcription and diarization job for an audio or video URL (YouTube, X, Instagram, TikTok). Returns a job ID that can be used to check status and download results.

Name Type Required Description

url string required The URL of the audio or video content to transcribe (e.g. YouTube, X, Instagram, TikTok link)

language string optional Language code for transcription (e.g. 'en', 'es', 'fr'). Defaults to auto-detection if not provided.

num_speakers integer optional Expected number of speakers in the audio. Helps improve diarization accuracy.

schema_version string optional Optional schema version to use for tool execution

tool_version string optional Optional tool version to use for execution

diarize_download_transcript Download the transcript output for a completed transcription job in JSON, TXT, SRT, or VTT format, including speaker diarization, segments, and word-level timestamps. 4 params ▾

Download the transcript output for a completed transcription job in JSON, TXT, SRT, or VTT format, including speaker diarization, segments, and word-level timestamps.

Name Type Required Description

job_id string required The unique ID of the completed transcription job

format string optional Output format for the transcript. Supported formats: 'json', 'txt', 'srt', 'vtt'.

schema_version string optional Optional schema version to use for tool execution

tool_version string optional Optional tool version to use for execution

diarize_get_job_status Retrieve the current status of a transcription job by its job ID. Returns job state (pending, processing, completed, failed), metadata, and an estimatedTime field (in seconds) indicating how long processing is expected to take. Use estimatedTime to determine polling frequency and max wait duration — for example, a 49-minute episode may have an estimatedTime of ~891s (~15 mins), so the agent should wait at least that long before giving up. 3 params ▾

Retrieve the current status of a transcription job by its job ID. Returns job state (pending, processing, completed, failed), metadata, and an estimatedTime field (in seconds) indicating how long processing is expected to take. Use estimatedTime to determine polling frequency and max wait duration — for example, a 49-minute episode may have an estimatedTime of ~891s (~15 mins), so the agent should wait at least that long before giving up.

Name Type Required Description

job_id string required The unique ID of the transcription job to check

schema_version string optional Optional schema version to use for tool execution

tool_version string optional Optional tool version to use for execution