How does the SDK handle user interruptions?

It uses AzureSemanticVadTurnDetection to detect speech patterns and silence, allowing the AI to naturally pause or respond based on configurable thresholds.

What is the Azure AI Voice Live SDK used for?

It is used to build real-time voice AI applications that require bidirectional communication, such as voice assistants and live speech-to-speech interaction using WebSockets.

Which AI models are compatible with this skill?

The skill supports GPT-4o-realtime-preview, GPT-4o-mini-realtime-preview, and Phi-4-mm-realtime for multimodal voice interactions.

Can I use function calling with real-time voice?

Yes, the SDK allows you to define VoiceLiveFunctionDefinitions and process function call arguments during the live session to fetch external data.

What audio format is recommended for Azure AI Voice Live?

The recommended format is 16-bit PCM (Pcm16) at 24kHz, mono channel, for both input and output audio streams.

Azure AI Voice Live for .NET

Name: Azure AI Voice Live for .NET
Author: microsoft

bymicrosoft

•

1,777

•

云基础设施

Builds real-time, bidirectional voice AI applications using the Azure AI Voice Live SDK for .NET.

This skill empowers developers to integrate sophisticated real-time voice capabilities into .NET applications using the Azure AI Voice Live SDK. It provides implementation patterns for establishing bidirectional WebSocket communications, configuring advanced voice models like GPT-4o-realtime, and managing complex interactions such as semantic voice activity detection and function calling. Ideal for building responsive voice assistants and conversational AI, this skill ensures best practices for session management, Entra ID authentication, and efficient audio processing within the Azure ecosystem.

主要功能

01Secure authentication using Microsoft Entra ID and DefaultAzureCredential

02Semantic Voice Activity Detection (VAD) for natural conversation flow

03Integrated function calling support within voice-enabled sessions

04Support for GPT-4o and Phi-4 multi-modal realtime models

051,777 GitHub stars

06Real-time bidirectional WebSocket communication for low-latency audio

使用场景

01Building real-time speech-to-speech translation and communication tools

02Implementing voice-controlled interfaces for hands-free .NET desktop or server applications

03Developing high-performance AI voice assistants and virtual agents

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add microsoft/skills azure-ai-voicelive-dotnet

For use in Claude.ai and ChatGPT

Download Skill