What is a SIP trunk?
A SIP trunk is a virtual connection that carries voice calls between a PBX or softswitch and the public telephone network or another carrier over an IP connection, replacing physical phone lines with a scalable, cloud-based alternative.
What is the difference between STUN and TURN?
STUN helps a device discover its own public IP address so two WebRTC peers can attempt a direct connection; it does not relay media. TURN relays media through a server when a direct connection is impossible, such as behind a symmetric NAT or strict firewall.
What is the difference between a Class 4 and Class 5 softswitch?
A Class 4 softswitch handles wholesale, carrier-to-carrier call routing at high volume. A Class 5 softswitch provides retail, end-user telephony features like voicemail, IVR, and conferencing.
VoIP & WebRTC Glossary
35 VoIP, WebRTC, and Voice AI terms defined precisely by CelloIP engineers — from core protocols (SIP, RTP, STUN/TURN/ICE) to servers (Asterisk, FreeSWITCH, Kamailio), architecture patterns (SFU/MCU, Class 4 vs Class 5 softswitch), and voice AI infrastructure (LiveKit, AudioSocket).
Core Protocols & Transport
SIP (Session Initiation Protocol)
SIP is the signalling protocol used to establish, modify, and terminate voice, video, and messaging sessions over IP networks. It handles call setup (INVITE, 180 Ringing, 200 OK, ACK) but not the media itself — RTP carries the actual audio and video once a SIP session is established.
Learn moreRTP (Real-time Transport Protocol)
RTP is the protocol that carries the actual audio and video packets of a call once a SIP session is set up. It handles sequencing, timestamping, and payload type identification but provides no encryption on its own — that's what SRTP adds.
SRTP (Secure RTP)
SRTP is RTP with encryption and message authentication added, protecting call audio and video from eavesdropping and tampering. Production VoIP and WebRTC deployments use SRTP by default; WebRTC in fact mandates DTLS-SRTP for every media stream.
WebRTC (Web Real-Time Communication)
WebRTC is a free, open-source, browser-native standard that enables real-time audio, video, and data communication without plugins or downloads. It underpins Google Meet, WhatsApp's calling feature, Discord, and thousands of custom communication apps.
Learn moreSTUN (Session Traversal Utilities for NAT)
STUN is a lightweight protocol a device uses to discover its own public IP address and port as seen from the internet, which is necessary because most devices sit behind a NAT (router). STUN does not relay media — it only helps two peers find each other's real network address.
TURN (Traversal Using Relays around NAT)
TURN is a media relay server used when direct peer-to-peer connection fails — typically behind symmetric NATs or restrictive corporate firewalls. Unlike STUN, all call audio and video flows through the TURN server, making it more bandwidth-intensive and the reason production WebRTC deployments need a dedicated coturn server.
Learn moreICE (Interactive Connectivity Establishment)
ICE is the framework that gathers all possible connection paths between two WebRTC peers — direct, via STUN-discovered address, or via TURN relay — and tests them to pick the best-working one automatically. Every WebRTC call runs an ICE negotiation before media starts flowing.
NAT Traversal
NAT traversal is the general problem of establishing a direct connection between two devices that both sit behind routers performing Network Address Translation, which hide their real IP addresses. STUN, TURN, and ICE together solve NAT traversal for real-time media.
Codec (G.711, Opus, VP8/H.264)
A codec compresses and decompresses audio or video for transmission. G.711 is the uncompressed baseline VoIP audio codec (64kbps, toll quality); Opus is the modern WebRTC standard offering better quality at lower bitrate and adaptive bitrate support; VP8 and H.264 are the dominant WebRTC video codecs.
DTMF (Dual-Tone Multi-Frequency)
DTMF is the tone signalling used when a caller presses phone keypad digits during a call — the technology behind IVR menu navigation ('Press 1 for sales'). In VoIP it's carried either as audio tones or, more reliably, as RFC 2833 out-of-band RTP events.
Diagram: SIP Call Setup Sequence
Diagram: WebRTC ICE Candidate Negotiation
Servers & Platforms
Asterisk
Asterisk is an open-source PBX and telephony engine used for SMB/enterprise phone systems, call centres, and IVR applications, with dialplan-based call routing and AGI/ARI interfaces for custom application logic.
Learn moreFreeSWITCH
FreeSWITCH is an open-source softswitch built for high-concurrency carrier and WebRTC workloads, with native mod_verto WebRTC support and an event-driven ESL (Event Socket Layer) interface for application integration.
Learn moreKamailio
Kamailio is a SIP proxy/router written in C, built for high-concurrency signalling — routing, load balancing, and SIP security at the network edge rather than handling media itself.
Learn moreOpenSIPS
OpenSIPS is a SIP proxy and session border controller platform used for carrier-grade routing, load balancing, and SIP security, positioned similarly to Kamailio with its own scripting and module ecosystem.
Learn morePJSIP
PJSIP is an open-source, multimedia communication library implementing SIP, used both as Asterisk's modern SIP channel driver (chan_pjsip) and as a standalone client library for building softphones and mobile SIP apps.
SBC (Session Border Controller)
An SBC sits at the edge of a VoIP network, enforcing security policy, protocol normalization, and encryption (TLS/SRTP) between untrusted external SIP trunks and the internal call platform. Kamailio and OpenSIPS are commonly deployed as software SBCs.
PBX (Private Branch Exchange)
A PBX is the phone system that routes calls within an organization and to the outside world. Traditional PBX hardware has been replaced almost entirely by software PBX platforms like Asterisk and FreeSWITCH running as VoIP softswitches.
Softphone
A softphone is a software application that turns a computer or mobile device into a SIP phone, registering to a PBX or SIP server and handling calls entirely in software rather than requiring a physical desk phone.
Learn moreArchitecture Patterns
SFU (Selective Forwarding Unit)
An SFU is a media server that receives one media stream from each call participant and selectively forwards only the streams each other participant needs, without mixing them. SFUs (LiveKit, Mediasoup, Janus) are required for any group video call beyond 3–4 participants because peer-to-peer WebRTC doesn't scale further.
MCU (Multipoint Control Unit)
An MCU mixes multiple participants' audio and video streams into a single combined stream before sending it to each participant, trading higher server CPU cost for lower client-side bandwidth and processing — the opposite trade-off from an SFU.
Class 4 Softswitch
A Class 4 softswitch routes wholesale, carrier-to-carrier voice traffic — it handles least-cost routing (LCR) and high call-per-second volume between carriers but has no end-user features like voicemail or IVR.
Learn moreClass 5 Softswitch
A Class 5 softswitch provides end-user telephony features — voicemail, call forwarding, IVR, conferencing — the retail layer of a phone network, as opposed to Class 4's pure wholesale call routing.
Learn moreCall Center & Business Logic
IVR (Interactive Voice Response)
An IVR is an automated phone menu system that lets callers navigate options via DTMF keypad presses or spoken input before optionally reaching a live agent — the foundation that AI voice bots now extend with LLM-driven natural conversation.
Learn moreACD (Automatic Call Distributor)
An ACD is the call center engine that routes incoming calls to available agents based on rules like skill, queue priority, or longest-idle-agent — the core logic behind every call center software platform.
Learn moreCDR (Call Detail Record)
A CDR is the log entry generated for every call — caller, callee, duration, timestamp, disposition, and cost — used for billing, analytics, and compliance across every VoIP platform.
Learn moreAGI (Asterisk Gateway Interface)
AGI is Asterisk's interface for running external scripts (Python, PHP, etc.) that control call flow synchronously via stdin/stdout — simple to use but blocking, which is why FastAGI and ARI exist for more demanding real-time applications.
ARI (Asterisk REST Interface)
ARI is Asterisk's modern, asynchronous REST and WebSocket API for building call-control applications, giving fine-grained control over channels, bridges, and media that AGI's synchronous model cannot provide — the preferred interface for real-time AI voice agents on Asterisk.
SIP Trunk
A SIP trunk is a virtual connection that carries voice calls between a PBX/softswitch and the public telephone network (PSTN) or another carrier over an IP connection, replacing physical phone lines with a cloud-based, scalable alternative.
Learn moreTelecom Business Models
MVNO (Mobile Virtual Network Operator)
An MVNO offers mobile phone service to consumers without owning the underlying wireless network infrastructure, instead leasing network capacity from a host mobile carrier and running its own billing, branding, and customer systems.
Learn moreMVNE (Mobile Virtual Network Enabler)
An MVNE provides the technical infrastructure — provisioning, billing platforms, OSS/BSS — that MVNOs use to operate, sitting between the host mobile network and the MVNO brand without directly serving end consumers itself.
Learn moreCPaaS (Communications Platform as a Service)
CPaaS is a cloud platform that lets developers embed voice, SMS, video, and messaging into applications via API, without building or operating the underlying telecom infrastructure — Twilio and Vonage are the best-known CPaaS vendors.
UCaaS (Unified Communications as a Service)
UCaaS bundles voice, video conferencing, chat, and collaboration tools into a single cloud-delivered business communication platform aimed at end-user organizations, distinct from CPaaS's developer-API focus.
Voice AI
Voice Bot / Conversational IVR
A voice bot is an AI phone agent that conducts natural spoken conversation using a speech-to-text → LLM → text-to-speech pipeline, replacing rigid DTMF-menu IVR with free-form conversational understanding.
Learn moreLiveKit
LiveKit is an Apache 2.0 open-source WebRTC SFU with official Flutter, React Native, iOS, Android, and Web SDKs, usable either self-hosted (no per-minute fees) or via LiveKit Cloud, and increasingly used as the real-time transport layer for AI voice agents.
Learn moreAudioSocket
AudioSocket is a TCP-based protocol that streams 8kHz 16-bit signed linear PCM audio in 320-byte frames between Asterisk and an external application, offering a non-blocking alternative to AGI for real-time AI voice agent integration.
Glossary FAQ: Commonly Confused Terms
What is the difference between SIP and WebRTC?
SIP is a signalling protocol traditionally used with the RTP media protocol in telephony networks and PBX systems. WebRTC is a browser-native standard that bundles its own signalling-agnostic media stack (ICE, DTLS-SRTP) directly into web and mobile apps. The two are often bridged together — a WebRTC browser client connects through a SIP-WebRTC gateway to reach a traditional SIP/PSTN network.
What is the difference between AGI and ARI in Asterisk?
AGI (Asterisk Gateway Interface) is a synchronous, script-based interface — your application blocks while Asterisk waits for a response. ARI (Asterisk REST Interface) is asynchronous, using REST calls and WebSocket events, giving far more granular real-time control over channels and media — the preferred choice for AI voice agents that need to interrupt or redirect a call mid-flow.
What is the difference between Class 4 and Class 5 softswitch?
A Class 4 softswitch handles wholesale, carrier-to-carrier call routing at high volume with least-cost routing logic and no end-user features. A Class 5 softswitch provides retail, end-user telephony features like voicemail, IVR, and conferencing. Many carrier platforms run both layers together.
What is the difference between STUN and TURN?
STUN helps a device discover its own public-facing IP address and port so two peers can attempt a direct connection; it does not relay media. TURN acts as a media relay of last resort when a direct connection is impossible (e.g., symmetric NAT or strict firewalls) — all audio/video then flows through the TURN server, making it more expensive to run at scale.
What is the difference between an SFU and an MCU?
An SFU forwards each participant's media stream selectively without mixing it — lower server CPU cost, higher client bandwidth. An MCU mixes all streams into one combined stream before sending it out — lower client bandwidth, much higher server CPU cost. Nearly all modern group video platforms (LiveKit, Zoom, Google Meet) use SFU architecture.
Explore These Technologies
Have a VoIP or WebRTC Project in Mind?
Talk to a CelloIP engineer about your architecture — free 30-minute technical consultation.