A traditional landline call carries your voice as a continuous electrical signal down a dedicated copper wire, the same way it has since the telephone was invented. VoIP does something fundamentally different: it samples your voice thousands of times a second, converts each sample into a number, and sends those numbers across the same internet connection your business already uses for everything else. This is the same basic idea as an MP3 file or a Spotify stream, applied to a live, two-way conversation instead of a pre-recorded track, with one added constraint: it must happen fast enough that a conversation still feels like a conversation.
The seven-step journey of a VoIP call
Every VoIP call, whether it's a $30-a-month cloud phone system or an enterprise deployment, goes through the same seven steps. None of them are visible to the people talking, and all of them happen in a fraction of a second.
- Capture: A microphone in the handset, headset or softphone converts sound waves into an analogue electrical signal, exactly as any telephone or microphone has always done.
- Digitisation: That analogue signal is sampled thousands of times per second and converted into a stream of digital numbers, a process called pulse-code modulation. This is the same principle used to digitise any analogue audio.
- Compression (the codec): A codec, short for coder-decoder, compresses that digital stream so it takes up far less bandwidth. Different codecs trade audio quality against bandwidth differently: G.711 preserves near-landline quality but uses more bandwidth, G.729 compresses harder for low-bandwidth connections, and Opus, increasingly common in modern systems, adjusts quality dynamically based on the connection. Your provider or phone system chooses this for you, but it's the single biggest factor in how a call actually sounds.
- Packetisation: The compressed audio is split into small chunks, typically representing about 20 milliseconds of speech each, and wrapped in a data packet using RTP (Real-time Transport Protocol). Each packet gets a sequence number and a timestamp, which turns out to matter a great deal in step six.
- Transmission: The packets are sent across the internet using UDP rather than TCP, the protocol most web traffic uses. This is a deliberate trade-off: UDP doesn't pause to confirm each packet arrived or resend lost ones, because for a live call, a briefly missing fragment of audio is far less disruptive than the delay that resending it would cause. Packets can and do take different routes across the internet to reach the same destination.
- Reassembly and the jitter buffer: Because packets can arrive slightly out of order or at uneven intervals, the receiving device holds them briefly in a jitter buffer, a small holding area that reorders packets using their sequence numbers and smooths out timing gaps before playback. A jitter buffer that's too small lets timing problems through as audio glitches; one that's too large adds noticeable delay to the conversation.
- Decompression and playback: The same codec used to compress the audio decompresses it back into a digital audio stream, which is converted back into an analogue signal and played through a speaker. Total elapsed time for all seven steps, capture to playback, is typically well under 150 milliseconds on a properly configured connection, comfortably within what a human conversation can tolerate without feeling delayed.
SIP and RTP: the two channels every VoIP call uses
One detail that trips up a lot of explanations of VoIP is treating it as a single protocol. In practice, a VoIP call runs on two separate systems working together, and understanding the difference explains a surprising number of real-world VoIP problems.
| Protocol | What it actually does | An analogy |
| SIP (Session Initiation Protocol) | Sets up, manages and ends the call: dialling, ringing, answering, hold, transfer and hang-up. It negotiates which codec both ends will use before any audio flows. | The receptionist who answers, announces the call, and connects you |
| RTP (Real-time Transport Protocol) | Carries the actual voice packets once the call is connected. This is the channel doing the work described in the seven steps above. | The phone line itself, once you're connected and talking |
This split matters practically: a call that connects and shows as "ringing" or "answered" correctly, but has no audio or one-way audio, is very often a firewall or router blocking or mishandling the RTP media stream while letting the SIP signalling through untouched. It's a specific, diagnosable fault, not a vague "VoIP is being flaky" problem, and knowing the split is often the fastest way to explain the symptom to whoever is troubleshooting it.
Why VoIP calls need special treatment on your network
A single VoIP call typically uses somewhere between 30 and 100 kilobits per second of bandwidth, depending on the code, which is tiny compared to most business internet connections. Bandwidth is rarely the actual problem. The real issue is that a phone call and, say, a large file upload is both competing for the same connection, and a standard network treats them identically. Quality of service (QoS) configuration, usually applied through DSCP tagging on your router or firewall, tells your network to prioritise voice packets ahead of other traffic, so a colleague's cloud backup running in the background doesn't introduce jitter into your call. This is a one-off configuration step, not an ongoing cost, and it's one of the most common gaps behind "our VoIP call quality is inconsistent" complaints that turn out to have nothing to do with the VoIP provider at all.
VoIP and calling 000: what Australian rules require
This is the part of VoIP that most explainers, including most provider marketing pages, skip over entirely, and it is genuinely important for any Australian business relying on a VoIP line.
Triple Zero (000) can be dialled from "certain" VoIP services, not automatically from all of them, and the distinction matters. A traditional landline draws its own power down the copper line and keeps working through a mains power outage; a VoIP service depends on both your internet connection and mains power at your premises, unless you have specifically planned around that with a UPS or a cellular failover connection. If either goes down, a VoIP line configured without backup simply cannot place an outbound call, including to 000.
There is also a location problem specific to VoIP. A traditional landline's physical address is fixed and known to the network. A VoIP service's emergency call carries whatever address is registered against that service in your provider's system, not your device's actual physical location, which is why keeping that registered address current after an office move or when adding a new site is a genuine safety requirement, not paperwork.
The regulatory backdrop has also tightened recently. The Telecommunications Legislation Amendment (Triple Zero Custodian and Emergency Calling Powers) Act 2025 commenced on 1 November 2025, establishing a Triple Zero Custodian within the federal communications department and giving the Australian Communications and Media Authority (ACMA) stronger powers to direct telecommunications providers, including new obligations to test emergency calling during network upgrades and maintenance and to communicate properly during outages. The changes followed high-profile failures, including an Optus network fault in 2023 and further emergency-calling outages investigated by the ACMA through late 2025, that left some callers unable to reach Triple Zero.
A short VoIP glossary
| Term | Meaning |
| Codec | Software that compresses voice into a smaller digital stream for transmission, and decompresses it on arrival |
| SIP | The protocol that sets up, manages, and ends a VoIP call (the signalling channel) |
| RTP | The protocol that carries the actual voice audio once a call is connected (the media channel) |
| Jitter buffer | A small holding area that reorders and smooths out voice packets before playback |
| Latency | The delay between speaking and being heard on the other end |
| Packet loss | Voice data that never arrives at all, causing gaps or dropouts |
| QoS (Quality of Service) | Network configuration that prioritises voice traffic ahead of other data on the same connection |
| PSTN | Public Switched Telephone Network: the traditional copper-based phone network VoIP is generally replacing |
For how this technology translates into an actual buying decision, choosing between a VoIP phone and a standard desk phone, or migrating an existing phone system to VoIP, the related guides linked here cover that ground in more practical detail than this explainer aims to.
Conclusion
VoIP explained plainly still comes down to this: your voice becomes data, that data travels across the internet in small, independently routed packets, and it arrives sounding like an ordinary phone call, provided the network underneath it is set up properly. Understanding the seven steps in between, and the two protocols working together to make them happen, turns "our VoIP is being flaky" into a specific, diagnosable problem instead of a vague complaint. And for any Australian business relying on VoIP as its only phone line, confirming Triple Zero calling and keeping your registered address current isn't an optional detail; it's a safety basic worth checking today rather than during an emergency.
