Self-hosted gateway that connects phone lines and chat channels to AI agents, using your own API keys.
Azure-Samples/voicelive-api-salescoach
A demo app that uses Azure Voice Live API to let sales professionals practice voice conversations with AI-powered virtual customers.
- Stars
- 150
- Forks
- 94
- Open issues
- 8
- Last commit
- 29 days ago
What is it?
Voice Live API Sales Coach is a sample application that shows how AI-driven voice training can support sales education. Users pick an industry-specific sales scenario, speak with an AI virtual customer in real time, and receive feedback on their performance.
The app is built on Azure AI services. Conversations run through the Azure Voice Live API for speech-to-speech interaction, a large language model analyzes the conversation afterward, and Azure Speech Services provides pronunciation and fluency assessment. The frontend is React with Fluent UI, and the backend is a Python Flask server that handles WebSocket communication.
The project includes a dev container, a build script, and an `azd up` deployment that provisions Azure AI and Speech resources and a Container App. It can use Azure OpenAI keys or Microsoft Entra ID authentication, and optionally connects to Foundry agents for scenario behavior.
Key features
- 01Real-time voice conversations with AI virtual customers via Azure Voice Live API
- 02Post-conversation performance analysis covering tone, content, needs assessment and objection handling
- 03Pronunciation and fluency assessment through Azure Speech Services
- 04Industry-specific sales scenarios with scoring metrics
- 05One-command deployment to Azure with azd
Use cases
- Training sales staff to handle objections and discovery questions through simulated calls
- Exploring how Azure Voice Live API and Speech services fit into an interactive voice application
- Using as a reference implementation for a Flask and React app with real-time voice
Quick start
- 1
Deploy to Azure with azd
azd up - 2
Install and build the app locally
./scripts/build.sh - 3
Start the server
cd backend && python src/app.py - 4
Open http://localhost:8000 in your browser
Things to know
- Requires an Azure subscription with Voice Live, Speech and a gpt-4.1-mini model deployment, plus enough model quota.
- Local development needs a filled-in .env file with Azure AI Foundry and Speech keys or endpoints.
Topics
FAQ
- What is voicelive-api-salescoach?
- A demo app that uses Azure Voice Live API to let sales professionals practice voice conversations with AI-powered virtual customers. Voice Live API Sales Coach is a sample application that shows how AI-driven voice training can support sales education. Users pick an industry-specific sales scenario, speak with an AI virtual customer in real time, and receive feedback on their performance.
- Is voicelive-api-salescoach free to use? What license does it have?
- voicelive-api-salescoach is open source under the MIT license. Check the license terms before using it in commercial products.
- Is voicelive-api-salescoach still maintained?
- Actively maintained. The last commit to voicelive-api-salescoach was on Sep 21, 2026.
- What can I use voicelive-api-salescoach for?
- Training sales staff to handle objections and discovery questions through simulated calls Exploring how Azure Voice Live API and Speech services fit into an interactive voice application Using as a reference implementation for a Flask and React app with real-time voice
- How do I install voicelive-api-salescoach?
- Deploy to Azure with azd: azd up Install and build the app locally: ./scripts/build.sh Start the server: cd backend && python src/app.py Open http://localhost:8000 in your browser
- Where can I get voicelive-api-salescoach?
- The source code and installation instructions are on GitHub: https://github.com/Azure-Samples/voicelive-api-salescoach
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
A 21-lesson Microsoft course teaching developers the fundamentals of building generative AI applications with Python and TypeScript.
Open-source document parser that converts PDF, Office, HTML and other files into structured data for generative AI pipelines.
A memory layer for AI agents and apps that stores and retrieves user, session, and agent context across conversations.
Turns a Git repository or local directory into a single prompt-friendly text digest for feeding code to LLMs.
A modular Agentic RAG tutorial and project built with LangGraph, featuring hierarchical indexing, conversation memory, and query clarification.