CalloraVoipSdk

Build your own voice product on a sovereign telephony core.

European B2B voice runtime for teams building calling, dialer, contact center, or voice AI products — with full technical control over telephony, media path, and intelligent decision logic.

Built for

  • PBX and UC vendors
  • Contact center software makers
  • Dialer and campaign tools
  • CRM/Sales automation with calling
  • Voicebot and AI agent platforms
  • Fraud, spam and scam detection systems

Current Status

Latest release: v4.14.0 on nuget.org — the version this documentation describes.

What each release changed, and why, is in the release notes; the full list of changes is in the changelog. Every release in the 4.x line has been additive — no public API has been removed or changed, which the tracked PublicApi.approved.txt baseline enforces on every build.

The table below is what matters here: which parts are mature, which are opt-in, and which are not implemented at all.

How to read the status column. Stable = mature, covered by the RFC-oriented test suite and by automated interop, and the intended production surface. Opt-in = shipped and tested, but off by default and not yet proven in production traffic — validate for your environment first. The production-proven NAT path is symmetric RTP (comedia), which needs no ICE or STUN. The SIP + RTP core is exercised by an automated interop suite against a real Asterisk (PJSIP) container in CI — calls, media, codecs, SRTP-SDES, DTMF, transfer, session timers, early media and TCP/TLS, plus a two-leg bridged call with byte-exact bidirectional media (currently all cases green, none skipped). The WebRTC path is validated in CI against real Chromium and Firefox and a real coturn relay (see the interop matrix). Known gaps and interop defects are tracked openly in the issue tracker.

Core (SIP + RTP):

Capability Status
SIP Register / Dial / Accept / Hangup ✅ Stable
Hold / Unhold / Blind + Attended Transfer ✅ Stable
REFER transfer progress subscription (RFC 3515 / 6665) ✅ Stable
Early media (RFC 3960): pre-answer receive-only media + DTMF in the early dialog ✅ Stable
RTP media transport ✅ Stable
SRTP + SRTCP media encryption (SDES, offer & answer; RFC 4568 / RFC 3711) ✅ Stable
DTLS-SRTP media encryption (RFC 5763, incl. AEAD-AES-GCM per RFC 7714) ✅ Stable
Adaptive jitter buffer ✅ Stable
Media cross-connect / bridge ✅ Stable
Per-call media tap (frame receivers/senders for bots and streaming) ✅ Stable
Module registry (client.Modules) as plugin extension point ✅ Stable
Configurable audio codec preference ✅ Stable
DTMF send/receive (RFC 4733) ✅ Stable
SIP MESSAGE send & receive (RFC 3428) ✅ Stable
SIP PUBLISH: event-state publish / refresh / modify / remove (RFC 3903) ✅ Stable
RTCP quality metrics (measured jitter, loss, round-trip time) ✅ Stable
Recording + Playback (WAV/MP3) ✅ Stable
Linux + Windows audio devices ✅ Stable
Runtime device hot-switch + controls ✅ Stable
Encoded video: send/receive, transport-cc bitrate recommendation, keyframe feedback (transport-only) ✅ Stable (single-stream)

WebRTC and NAT traversal:

Capability Status
WebRTC facade: peer connections, SDK-driven signalling, W3C tracks, media taps (transport-only) ✅ Stable — browser-validated (Chromium + Firefox, both roles)
WebRTC video repair & congestion control: NACK/PLI/FIR, RTX, transport-cc, getStats ✅ Stable
Send-side simulcast (RFC 8853 / 8852) ✅ Stable
Data channels (SCTP) 🚫 Not included
Self-hostable STUN / TURN server (RFC 5389 / 5766 / 8656) ✅ Stable — verified against coturn (UDP relay only)
ICE for NAT traversal (RFC 8445/7675: role + tie-breaker, check-list FSM, nomination, inbound/triggered checks, consent freshness, restart incl. RestartIceAsync) ⚙️ Opt-in — off by default, unproven in production trunks
ICE-TCP candidates (RFC 6544) 🚫 Not included (deliberate — trunk calls use symmetric RTP)
Backend/API for signed plugin marketplace + tenant entitlements 📋 Roadmap

Multi-party / SFU enablement (4.7, WebRTC guide):

All of these are additive and transport-only — a peer that uses none of them negotiates byte-identical SDP to 4.6. The SDK stays a peer: it forwards, it does not mix or transcode.

Capability Status
Multiple video tracks over one BUNDLE (AddVideoTrack, own m=video + SSRC per track, RemoteTrack.Mid) ✅ Stable
Multiple audio tracks over one BUNDLE (AddAudioTrack, per-participant a=msid) ✅ Stable
Mid-call renegotiation (RFC 8829): apply the track delta live, no transport / DTLS / ICE / SRTP rebuild ✅ Stable
Signalling state observation (SignalingState, SignalingStateChanged — W3C RTCSignalingState) ✅ Stable
Receive-side simulcast demux (RFC 8853 / 8852): per-RID reassembly, layer id on EncodedFrame.Rid ✅ Stable (forwarding-only — layer choice is the app's)
Per-peer send-bitrate recommendation from transport-cc (RecommendedOutgoingBitrateBps, RecommendedBitrateChanged) ✅ Stable
On-demand key-frame request per track (RequestVideoKeyFrameAsync(mid), RFC 4585 §6.3.1 PLI) ✅ Stable
ICE restart on a connected peer (CreateIceRestartOfferAsync, re-offer rotating the ICE ufrag) ✅ Stable
TURN relay over TCP/TLS (stream relay, RFC 8656 §12 ChannelData) ✅ Available (unit-proven; browser/real-server interop tracked in the matrix)
Audio mixing / transcoding (a real conference server) 🚫 Not included — out of scope by design

Choose your integration depth

CalloraVoipSdk uses a progressive API. Start with managed workflows for registration, dialing, default audio, playback and recording. When the product needs more control, the same call remains available through typed IPhoneLine/ICall contracts, encoded media receivers/senders, cross-connect, custom devices, telemetry and modules.

The boundary is deliberate: supported call, media and extension seams are public, while transport/parser implementation types and arbitrary wire mutation remain internal. Choose the right depth →

SDK Structure

CalloraVoipSdk.Core — Sovereign calling foundation

Clean DDD architecture: Domain → Application → Infrastructure → public VoipClient facade. No vendor lock-in. Full protocol stack owned in-house (SIP, RTP, SRTP, DTLS-SRTP, SDP, STUN/TURN client and server) — no external SIP/RTP/ICE library.

Commercial plugins (private feed, licensed separately — in development)

The SDK core is open and free. Advanced capabilities ship as paid plugins on a private feed, built on the public module registry and media-tap contract:

  • Callora.Realtime — bridge call audio to realtime AI APIs (e.g. OpenAI Realtime) with pacing, backpressure and barge-in; the foundation for AI voice agents
  • Callora.WebSocket — raw call-audio streaming over WebSocket
  • Callora.Privacy — redaction, consent management, policy gates, audit trail
  • Callora.Risk — spam/scam signals, call risk screening, PBX abuse prevention
  • Callora.Intelligence — AMD, sentiment, transcription, local model integration

Interested in early access? Contact info@bechstein.digital.

Quickstart

using var client = new VoipClient(new VoipConfiguration
{
    LoggerFactory = loggerFactory,
    UserAgent = "MySoftphone/1.0"
});

var connectResult = await client.ConnectAsync(new SipAccount
{
    Username = "1001",
    Password = "secret",
    SipServer = "pbx.example.com"
});

if (!connectResult.IsSuccess || connectResult.Line is null)
    throw new InvalidOperationException($"Connect failed: {connectResult.Status}");

var dialResult = await client.DialAndWaitUntilConnectedAsync(
    connectResult.Line,
    "sip:1002@pbx.example.com");

if (!dialResult.IsSuccess || dialResult.Call is null)
    throw new InvalidOperationException($"Dial failed: {dialResult.Status}");

await client.AttachDefaultAudioAsync(dialResult.Call);
await dialResult.Call.HangupAsync();

→ Getting Started · Progressive API · Core Concepts · Guides · WebRTC · Interop · Production · Capacity

Edit this page