# Building Enterprise Realtime Voice Agents from Scratch: A Technical Tutorial

arXiv: https://arxiv.org/abs/2603.05413 · HTML: https://mlsanity.com/paper/2603.05413/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 5 Mar 2026.

## Built on by (most cited first, 4 of 4)

- [Enabling Proactive Spoken Turns via a Generalized Style-Aware Full-Duplex Framework](https://mlsanity.com/paper/2608.28630/) (2026)
- [When Patients Cut In: Extending Clinical Conversational AI Safety to Interruptions](https://mlsanity.com/paper/2608.29241/) (2026)
- [Evaluating Real-Time Voice Agents: From Component Quality to Grounded Outcomes](https://mlsanity.com/paper/2609.30798/) (2026)
- [Following the Flow the Law Requires: Training Voice Agents to Transfer to Unseen Regulated Procedures in Debt Collection](https://mlsanity.com/paper/2610.18881/) (2026)

## Abstract (arXiv)

We present a technical tutorial for building enterprise-grade realtime voice agents from first principles. While end-to-end speech-to-speech models may ultimately provide the best latency for voice agents, fully self-hosted end-to-end solutions are not yet available. We evaluate the closest candidate, Qwen3-Omni, across three configurations: its cloud-only DashScope Realtime API achieves ~702ms audio-to-audio latency with streaming, but is not self-hostable; its local vLLM deployment supports only the Thinker (text generation from audio, 516ms), not the Talker (audio synthesis); and its local Transformers deployment runs the full pipeline but at ~146s -- far too slow for realtime. The cascaded streaming pipeline (STT rightarrow LLM rightarrow TTS) therefore remains the practical architecture for self-hosted realtime voice agents, and the focus of this tutorial. We build a complete voice agent using Deepgram (streaming STT), vLLM-served LLMs with function calling (streaming text generation), and ElevenLabs (streaming TTS), achieving a measured time-to-first-audio of 755ms (best case 729ms) with full function calling support. We release the full codebase as a 9-chapter progressive tutorial with working, tested code for every component.