Skip to main content

​
Overview

Not using one of our pre-built integrations? No problem. Send call data from any voice AI platform or custom application — just provide an audio file and Roark handles the rest: transcription, speech analysis, sentiment detection, and metric collection.

​
Send a Call

Install the Node.js SDK:
npm install @roarkanalytics/sdk

import Roark from '@roarkanalytics/sdk'

const client = new Roark({
  bearerToken: process.env.ROARK_API_BEARER_TOKEN,
})

const call = await client.call.create({
  recordingUrl: 'https://your-storage.com/call-recording.mp3',
  startedAt: '2024-01-15T10:00:00Z',
  interfaceType: 'PHONE',
  callDirection: 'INBOUND',
  agent: {
    name: 'Support Agent',
    customId: 'custom-agent-1',
  },
  customer: {
    phoneNumberE164: '+1234567890',
  },
  properties: {
    department: 'sales',
    region: 'us-east',
  },
})
Once a call is created, it appears in Call History and is automatically transcribed and analyzed.

​
What You Can Send

DataDescription
Audio fileMP3, WAV, M4A, or FLAC — via public URL or base64
MetadataAgent info, customer details, call direction, interface type
Custom propertiesArbitrary key-value pairs for filtering and segmentation
Tool callsFunction/tool invocations made during the conversation
Pre-transcribed textIf you already have a transcript, send it along

​
Audio Requirements

SpecRequirement
FormatMP3, WAV, M4A, or FLAC
Sample Rate8kHz or higher
ChannelsMono or stereo
DurationUp to 2 hours
File SizeMax 500MB
AccessPublicly accessible URL or base64 encoded

​
What Happens Next

Once a call is ingested, Roark automatically:
  1. Transcribes and analyzes the conversation — speech patterns, sentiment, 64+ emotions, interruptions, and more
  2. Runs metric policies — any active metric policies collect metrics automatically (LLM evaluations, compliance checks, custom KPIs)
  3. Makes it searchable — the call appears in Call History with full filtering by properties, agent, duration, and more

​
Use With Simulations

Agents created via the API can also be used in simulations. When you send calls with an agent.customId, Roark creates or matches an agent record that you can then target in simulation run plans — useful for testing custom-built voice agents that aren’t on a pre-built platform.

​
MCP Server

If you use an AI-powered development environment, the Roark MCP Server gives your AI assistant direct access to the Roark API — including creating calls, querying metrics, and managing agents — without writing integration code manually.

​
Getting Started

1

Generate an API Key

Create an API key in your Roark dashboard
2

Send Your First Call

Upload an audio file using the SDK or REST API above
3

Configure Metric Policies

Set up metric policies to automatically collect metrics on every call
4

Explore Your Data

Browse calls in Call History, build reports, and create dashboards

​
Resources

Enable Tracing

Send OpenTelemetry traces from any platform

Node.js SDK

Full SDK guide and reference

Python SDK

Full SDK guide and reference

API Reference

Complete REST API documentation