Voice Agent API

Stream audio in, get audio back. We handle the rest so you can focus on your product.

Click to start a voice conversation with our AI support agent.

Agent

Ask questions about our products, APIs, and documentation to experience real-time Voice AI in action.

Purpose-Built for Speech

The most accurate voice agent

Universal-3 Pro gets the details right — emails, phone numbers, order IDs, and names — so your voice agent can actually complete customer tasks.

0:00 / 0:51

AssemblyAI Voice Agent APIAccurate transcript

AgentThanks for calling the prescription refill line. This is Priya. Can I have your date of birth and the RX number on the bottle?

CallerDate of birth is 10/10/51 and the prescription is RX-7704132 It's Metropolol 80mg

AgentGot it. I'm also seeing a standing order from Dr. Chen for epinephrine, 0.25mg 1:1,000 IM Do you want that refill too?

CallerYes, please. And can you send it to the address on file? 10631 Northeast Knott Street, Portland, Oregon 97220

AgentConfirmed. I'll route it to your MyChart, and you'll get a text at (971) 235-7292 when it's all ready, around 08:30 tomorrow.

Deepgram Voice Agent API

Agentthanks for calling the prescription refill line this is priya can i have your date of birth and the rx number on the bottle

Callerdate of birth is october tenth fifty one and the prescription is rx seven seven zero four one three two it's metoprolol eighty milligrams

Agentgot it i'm also seeing a standing order from doctor chen for epinephrine zero point two five milligrams one to one thousand i'm do you want that refill too

Calleryes please and can you send it to the address on file one zero six three one northeast knot street portland oregon ninety seven thousand two hundred twenty

Agentconfirmed i'll route it to your mychart and you'll get a text at nine seven one two three five seven two nine two when it's all ready around eight thirty tomorrow

Purpose built

Your agent is only as good as what it actually hears

Names, accents, medical terms — we get it right where others approximate.

Industry-leading accuracy

Lowest word error rate on real-world audio. Email addresses, phone numbers, and entity names transcribed correctly so the LLM responds to what was actually said.

Turn detection that feels right

Knows when you're done talking vs. pausing to think. Doesn't cut you off mid-sentence, and stops listening when you interrupt.

~1 second response time

End-to-end latency fast enough that conversations flow naturally. You say something, the agent responds, no awkward pauses.

Natural voice generation

Purpose-built TTS voices tuned for conversation, not narration. Prosody, pacing, and intonation built for real-time dialogue.

Session resumption

Reconnect within 30 seconds if the WebSocket drops. Context preserved, conversation continues where it left off.

6 languages supported

English, Spanish, French, German, Italian, and Portuguese, with the same accuracy across all six.

Compare APIs

Model

AssemblyAI Voice Agent API

OpenAI Realtime API

Deepgram Voice Agent API

Price

$4.50/hr

$18.00/hr

$4.50/hr

ASR model

Universal-3 Pro

Gpt-realtime

Deepgram Nova-3

End-to-end latency

~1 second

~1–1.5 seconds

Ease of deployment

~6 event types

30+ event types

Moderate

Turn detection

Speech-aware VAD

Built-in

Basic

Live mid-conv config

No reconnect needed

Limited

—

Session resumption

30s reconnect window

—

Tool calling

JSON Schema

Included

Billing model

Flat hourly rate

Per-token audio

Component-based

Voice experience

Conversations that flow naturally

The most accurate voice agents on the market

Clean interruption handling + turn detection

~1 second response time

We own the stack, so every upgrade ships together

Developer experience

The fastest path to a working voice agent

Standard JSON API, no SDKs

Update prompts, voice, tools mid-call

Tool calling with JSON Schema

30s reconnect, context preserved

Use cases

Invisible infrastructure for your voice product

Full control over conversation flow, tools, and behavior. Your customers feel like you built it, because you did.

Customer Support

Agents that resolve tickets, look up accounts, and escalate when needed. Accurate enough to understand any caller on the first try.

Outbound Sales

SDR and sales agents that qualify leads, book meetings, and handle objections. Natural pacing and turn detection that doesn't sound like a robot.

Clinical Workflows

Voice interfaces for patient intake, triage, and documentation. Clinical-grade accuracy on medical terminology, with HIPAA compliance built in.

Scheduling and Intake

Receptionists and front-desk agents for appointments, intake forms, and routing. Handles names, phone numbers, and dates without misfires.

Phone Agents

Voice agents for inbound and outbound calls. Works with Twilio, LiveKit, and any telephony provider out of the box.

Voice Assistants

Voice interfaces inside your existing app. Give users a faster way to query data, trigger workflows, and navigate complex software.

Playground

We're not playing around, but you can

Put our Voice AI models to the test in our no-code playground.

Try it out

Common questions

: Voice Agent API is a single connection that handles speech-to-text, LLM routing, and voice generation for production voice agents. It's built end-to-end on AssemblyAI's Universal-3 Pro ASR with ~1 second end-to-end latency, speech-aware VAD turn detection, JSON Schema tool calling, and a 30-second reconnect window. Billing is flat at $4.50/hr.
: Yes. Sign up for a free account in the dashboard and get API credit to start building — no credit card required.
: Yes, we offer volume discounts for customers running large amounts of voice-agent traffic. Contact our sales team to discuss pricing that matches your workload.
: Voice Agent API runs at ~1 second end-to-end latency from the user finishing a turn to the agent starting to speak — fast enough that conversations feel natural.
: Voice Agent API is a single flat hourly rate of $4.50/hr that covers speech-to-text, LLM reasoning, and voice generation on one connection. You pay per second of session duration.
: Reach our sales team via the Contact us page, or chat with us directly in Discord. Enterprise customers get a dedicated account team.
: Voice Agent API supports English, Spanish, French, German, Italian, and Portuguese, with the same accuracy across all six.
: If a WebSocket drops, you have 30 seconds to reconnect. Conversation context is preserved on our side, so the agent picks up exactly where it left off — the user doesn't have to repeat themselves.