AI Free course · Certificate included
Voice AI & Conversational Agents
Design, build, test and launch voice agents with ElevenLabs, OpenAI Realtime and telephony, in English, Arabic and Urdu
- Intermediate
- 7 h 41 min
- 17 lessons in 7 modules
- 2 h 26 min of video lectures
- Updated Sep 2026
About this course
Voice agents now answer phones, book appointments, qualify leads and narrate content. This intermediate, hands-on course shows you how to build them properly. You will learn cascaded (ASR to LLM to TTS) and speech-to-speech architectures, latency budgets, turn-taking and barge-in; choose and design voices with ElevenLabs; build agents on ElevenAgents with tools, knowledge and tests; and use code-first options such as the OpenAI Realtime API, Gemini Live, LiveKit Agents and Pipecat. You will connect telephony with Twilio and SIP, design conversations and prompts for the ear, handle Urdu, Arabic and English code-switching, and apply consent, disclosure, recording and calling rules. Finally you will evaluate, monitor and cost voice agents, produce narration and course lectures, and build an appointment-booking agent as your capstone.
Tools you’ll use
- ElevenLabs
- ElevenAgents
- OpenAI Realtime API
- OpenAI Agents SDK
- Gemini Live API
- Twilio
- LiveKit Agents
- Pipecat
- FastAPI
Skills
- Voice agent design
- Conversational AI
- Text-to-speech
- Speech recognition
- Telephony integration
- Voice UX
- AI evaluation
What you’ll be able to do
- Choose between cascaded and speech-to-speech voice architectures for a given use case
- Budget and reduce voice-to-voice latency and tune turn-taking, barge-in and backchannels
- Build a voice agent on ElevenAgents with webhook tools, knowledge base, workflows and tests
- Build a browser voice agent with the OpenAI Realtime API and Agents SDK using ephemeral keys
- Connect agents to phone networks via Twilio or SIP with warm transfers and voicemail handling
- Design voice UX and prompts, including Urdu, Arabic and English code-switching
- Apply voice cloning consent, AI disclosure, recording and outbound calling rules
- Evaluate, monitor and cost voice agents and ship an appointment-booking agent
Recommended first
Curriculum
Syllabus
- Modules
- 7
- Lessons
- 17
- Reading time
- 4.5 h
- Assessment questions
- 25
How modern voice agents work, the difference between cascaded and speech-to-speech architectures, and why latency and turn-taking decide whether callers stay on the line.
- How voice agents work: cascaded pipelines vs speech-to-speechVideo lecture, 9′15 min
- Latency budgets, turn-taking and barge-inVideo lecture, 9′15 min
ElevenLabs voices and TTS models, building agents on ElevenAgents, and code-first options: OpenAI Realtime, Gemini Live and open-source frameworks.
- ElevenLabs voices: TTS models, voice design and cloningVideo lecture, 9′15 min
- Building an agent with ElevenAgentsVideo lecture, 9′18 min
- OpenAI Realtime API, Gemini Live and open-source stacksVideo lecture, 9′16 min
Connect voice agents to phone networks with Twilio and SIP, handle inbound and outbound calls, transfers, voicemail and keypad input.
- Telephony: phone numbers, SIP trunks and TwilioVideo lecture, 9′17 min
Voice UX, prompt design for spoken agents, tools and knowledge that work under latency, and multilingual design for Urdu, Arabic and English.
- Voice UX: designing conversations people enjoyVideo lecture, 9′15 min
- Prompt design for voice agentsVideo lecture, 9′15 min
- Tools and knowledge in voice agentsVideo lecture, 8′16 min
- Multilingual voice: Urdu, Arabic and EnglishVideo lecture, 8′16 min
Voice cloning consent and ethics, AI disclosure, recording consent, calling rules and data protection for voice agents across the EU, UK, US, Gulf and Pakistan.
- Voice cloning: consent, ethics and safeguardsVideo lecture, 9′15 min
- AI disclosure, recording consent and calling rulesVideo lecture, 9′16 min
Evaluate voice agents before and after launch, monitor them in production, control cost and scale safely.
- Evaluating and QA-testing voice agentsVideo lecture, 8′16 min
- Monitoring, cost control and scalingVideo lecture, 8′15 min
Apply voice AI to sales qualification, support, bookings and produced audio such as narration, podcasts and course lectures, then build an appointment-booking agent end to end.
- Use cases: sales qualification, support and bookingsVideo lecture, 8′15 min
- Produced audio: narration, podcasts and course lecturesVideo lecture, 8′15 min
- Capstone: build an appointment-booking voice agentVideo lecture, 8′25 min
- Final assessment
Your certificate
Finish with a credential anyone can check
Earn the Certified Voice AI Agent Builder badge: The holder can design, build, test and operate voice AI agents: choosing architectures, managing latency and turn-taking, building on ElevenAgents and realtime APIs, connecting telephony, designing multilingual conversations and prompts, applying consent, disclosure and calling rules, and evaluating, monitoring and costing agents in production.
Completed all lessons and scored at least 80% on the final assessment.
- A public verification page
- A PDF certificate to download
- An Open Badge you can share
- One click to your LinkedIn profile
Final assessment
- 25questions drawn from a larger pool
- 40 mintime limit
- 80%pass mark
- 3attempts per 24 hours
Start learning today. It’s free.
Every lesson is free to read. A free account saves your progress, unlocks the final assessment and issues your certificate.