VoipNet.Audio
1.2.0
dotnet add package VoipNet.Audio --version 1.2.0
NuGet\Install-Package VoipNet.Audio -Version 1.2.0
<PackageReference Include="VoipNet.Audio" Version="1.2.0" />
<PackageVersion Include="VoipNet.Audio" Version="1.2.0" />
<PackageReference Include="VoipNet.Audio" />
paket add VoipNet.Audio --version 1.2.0
#r "nuget: VoipNet.Audio, 1.2.0"
#:package VoipNet.Audio@1.2.0
#addin nuget:?package=VoipNet.Audio&version=1.2.0
#tool nuget:?package=VoipNet.Audio&version=1.2.0
Voip.NET
A Rust-powered, .NET-friendly VoIP + AI platform. SIP, RTP and SRTP from a native engine; calls, conversations and contact-centre building blocks from a modern .NET 10 API; speech and language models wired straight into the audio path.
Made by Gravicode Studios, led by Kang Fadhil. ๐ฎ๐ฉ Baca dalam Bahasa Indonesia: README.id.md

await using var client = new VoipClient(new VoipClientOptions
{
Domain = "pbx.example.com", Username = "1001", Password = "secret", RegisterOnStart = true,
});
await client.StartAsync();
var call = await client.CallAsync("sip:1002@pbx.example.com");
call.SendAudio(pcm16k, sampleRate: 16000);
Console.WriteLine($"{call.Codec} ยท MOS {call.GetStatistics().Mos:0.0}");
await call.HangupAsync();
Why Voip.NET
- Native where it matters. The SIP user agent, SDP negotiation, RTP/RTCP, SRTP, the adaptive jitter buffer, codecs and ICE live in Rust. Audio crosses into .NET through zero-copy callbacks.
- .NET where you work.
async/await, events,IAsyncEnumerableaudio streams, dependency injection,System.Diagnostics.Metrics. - AI in the call, not beside it. Speech-to-text, any
IChatClientmodel, text-to-speech and barge-in run as one loop on the live call. Realtime speech-to-speech models are supported too. - Contact centre included. IVR flows with AI hand-off, queues with skill routing, bridged agent legs, recording, analytics and CRM tools.
Packages
| Package | What it gives you |
|---|---|
VoipNet.Core |
VoipClient, VoipCall, conferences, DTMF, SRTP and DTLS-SRTP, TLS and WebSocket signaling, diagnostics (pcap, RTP analyser, metrics). Contains the native engine. |
VoipNet.Audio |
Microphone and speakers (OpenAL, cross-platform), WAV/MP3 call recording, resampling, voice activity detection, tones. |
VoipNet.AI |
Chat clients for OpenAI, Azure OpenAI, DeepSeek/OpenAI-compatible, Anthropic Claude and Google Gemini; speech providers; VoiceAgent; RealtimeVoiceAgent. |
VoipNet.Enterprise |
IVR builder and runner, CallCenterService (queues, agents, routing), RecordingService, CRM tool set. |
VoipNet.Cli |
voipnet .NET tool: SIP ping/register/call/listen/message and RTP analysis. |
Features
| Area | Status |
|---|---|
| SIP: REGISTER (digest auth, refresh, NAT keep-alive), INVITE/ACK/BYE/CANCEL, re-INVITE hold, REFER blind and attended (Replaces), OPTIONS, INFO, MESSAGE, NOTIFY | โ |
Transports: UDP, TCP, TLS (certificate pinning), WebSocket ws/wss (RFC 7118) |
โ |
| RTP/RTCP with adaptive jitter buffer, packet-loss concealment, symmetric RTP | โ |
| Codecs: Opus (48 kHz, in-band FEC), G.722, G.711 ฮผ-law/A-law, L16 (native) ยท G.729, SILK, Speex, H.264, VP8, VP9 (negotiated as pass-through) | โ / pass-through |
| DTMF: RFC 4733, SIP INFO, in-band generation and detection | โ |
| SRTP: AES-CM-128-HMAC-SHA1-80, AEAD-AES-128/256-GCM; keys via SDES or DTLS-SRTP | โ |
| WebRTC browsers calling SIP (SIP over WebSocket, ICE, DTLS-SRTP), verified with Edge and Firefox | โ ยท data channels planned |
| ICE (RFC 8445): candidate pairs, nomination, role conflicts, peer-reflexive candidates, trickle ICE, ICE restart, consent freshness ยท STUN, TURN | โ |
| Conferencing with mix-minus | โ |
| Recording WAV/MP3, stereo or mono | โ |
| LLMs: OpenAI, Azure OpenAI, DeepSeek & compatible servers, Anthropic, Gemini โ streaming and tool calling | โ |
| STT: Deepgram (streaming), OpenAI, Google Cloud, ElevenLabs, ElBruno.Realtime ยท TTS: ElevenLabs, OpenAI, Google Cloud, Amazon Polly, ElBruno.Realtime | โ ยท Amazon Transcribe planned |
| Voice agent: barge-in, sentence streaming, call-control tools, conversation memory, hand-off ยท Realtime speech-to-speech agent | โ |
| IVR builder, queues & skill routing, supervisor listen-in, recording service, metrics, CRM tools | โ |
Diagnostics: SIP to pcap, RTP stream analyser, dotnet-counters metrics |
โ |
| Platforms: Windows x64 built and tested; Linux/macOS build from source | โ / see building |
The full list, including what is planned, lives in PLAN.md and Progress.md.
Samples
| Sample | Kind | Shows |
|---|---|---|
| VoipNet.Softphone | Avalonia desktop | Dial pad, live call with audio traces, hold/mute/transfer/record, AI call summaries, built-in demo lines. |
| VoipNet.Gallery | Avalonia desktop | Every SDK feature as a live demo with its C#. |
| VoipNet.CallCenter | Blazor Server | Wallboard: queues, agents, live call quality, recordings, AI supervisor. |
| VoipNet.IvrStudio | Blazor Server | Design an IVR, call it from the browser, hand the caller to an AI agent. |
| VoipNet.WebPhone | Blazor Server | WebRTC gateway: the browser calls over SIP WebSocket and DTLS-SRTP, bridged to a SIP phone over UDP. |
| VoipNet.RealtimeAgent | Console | An AI agent that answers SIP calls (pipeline or realtime model). |
![]() |
![]() |
![]() |
![]() |

Quick start
# Requirements: .NET 10 SDK, Rust 1.80+ (only to build the engine from source)
git clone https://github.com/DotNetVibeCoderz/Vibe_Communication && cd Vibe_Communication/VoipNet
./build/build.ps1 # or ./build/build.sh on Linux/macOS
dotnet run --project samples/VoipNet.Softphone
Then read:
- Getting started
- Architecture
- Calls and media
- AI: models, speech and agents
- Contact centre: IVR, queues, recording, CRM
- Tools, diagnostics and samples
- Configuration reference
- Building and packaging
Credits
Voip.NET is made by Gravicode Studios, led by Kang Fadhil. Licensed under the MIT License. Sample apps use IBM Plex (SIL OFL) and Bricolage Grotesque (SIL OFL).
| Product | Versions Compatible and additional computed target framework versions. |
|---|---|
| .NET | net10.0 is compatible. net10.0-android was computed. net10.0-browser was computed. net10.0-ios was computed. net10.0-maccatalyst was computed. net10.0-macos was computed. net10.0-tvos was computed. net10.0-windows was computed. |
-
net10.0
- Microsoft.Extensions.DependencyInjection.Abstractions (>= 10.0.12)
- Microsoft.Extensions.Hosting.Abstractions (>= 10.0.12)
- Microsoft.Extensions.Logging.Abstractions (>= 10.0.12)
- Microsoft.Extensions.Options (>= 10.0.12)
- NAudio.Lame (>= 2.1.0)
- Silk.NET.OpenAL (>= 2.23.0)
- Silk.NET.OpenAL.Extensions.Enumeration (>= 2.23.0)
- Silk.NET.OpenAL.Extensions.EXT (>= 2.23.0)
- Silk.NET.OpenAL.Soft.Native (>= 1.23.1)
- VoipNet.Core (>= 1.2.0)
NuGet packages (2)
Showing the top 2 NuGet packages that depend on VoipNet.Audio:
| Package | Downloads |
|---|---|
|
VoipNet.AI
Voip.NET AI โ turn calls into conversations: streaming speech to text, large language models (OpenAI, Anthropic, Gemini, OpenAI-compatible), text to speech, barge-in and tool calling. |
|
|
VoipNet.Enterprise
Voip.NET Enterprise โ IVR flows with AI, call queues and agent routing, recording, analytics and CRM tools for contact centres. |
GitHub repositories
This package is not used by any popular GitHub repositories.



