Browser calling with WebRTC: webphones, softphones and IP phones

What browser calling with WebRTC actually means

WebRTC (Web Real-Time Communication) is an open standard from the W3C and the IETF that lets a web page use your microphone, speakers and network to exchange audio and video in real time. It ships inside every modern browser, including Chrome, Edge, Firefox and Safari, so there are no plugins or downloads involved.

For a business phone system, that means the phone itself can be a web page. Your agent opens the browser, signs in, and the phone built into the page, usually called a webphone, connects to the PBX. From there they answer calls, dial out, put callers on hold and transfer, just as they would on a desk phone.

The most obvious benefit is how little it takes: any computer with an up-to-date browser becomes a workstation, in the office or at home. Remote work, onboarding a new hire or moving desks no longer means shipping hardware or installing software. If you want the bigger picture of how an internet-based phone system works, start with our guide to what a VoIP phone system is.

How WebRTC works: microphone access, encryption and NAT traversal

A few moving parts sit behind every browser call. Knowing them helps, because they explain most of the problems you will run into day to day.

The first is microphone access. The page asks the browser for it through the getUserMedia API, and the browser shows the user a permission prompt. If the user declines, or blocked the site at some point in the past, the webphone cannot send any audio. Browsers only grant this access to pages served over HTTPS, so a webphone on an unencrypted page simply will not work.

• Encryption is mandatory: WebRTC media is always encrypted with SRTP, with keys negotiated over DTLS. It is not a setting someone can forget to switch on.
• Audio codecs: the standard requires browsers to support at least Opus and G.711. Opus is the usual choice between browsers because it adapts to the available bandwidth and copes well with less reliable networks.
• Signaling: WebRTC does not define how a call is set up. Phone systems typically use SIP carried over a secure WebSocket, or a protocol of their own.
• NAT traversal: the ICE protocol works out the best path between the browser and the server, helped by STUN servers, which discover your public address, and TURN servers, which relay the audio when a firewall blocks a direct path.

From the browser to the phone network: what the cloud PBX does

Your browser speaks WebRTC; the phone network does not. VoIP carriers use SIP for signaling and RTP for audio, often with codecs such as G.711. Something has to sit in between, and that is a gateway: a part of the PBX that receives the WebRTC call from the browser, removes the DTLS-SRTP layer, converts the codec if needed, and hands it to the PBX as an ordinary SIP call.

You never see this step. A webphone in the browser, a softphone on a laptop and an IP phone on a desk are simply three ways of using the same extension. The PBX applies the same queues, business hours, transfers and recordings whichever device the person is on.

With a cloud PBX, the gateway, SIP servers and TURN servers are run by the provider. Nothing needs installing on your premises, but you do depend on the quality of your connection to those servers, which is why the network matters more than the device for call quality. How this compares with an on-site system is covered in cloud PBX vs on-premise PBX.

Webphone, softphone or IP desk phone: which one to choose

No single device is best for everyone. It comes down to how each person works, and most businesses end up with a mix, sometimes with the same person using one at the office and another at home.

Setting up softphones and IP phones: SIP credentials and auto-provisioning

A softphone such as Zoiper, MicroSIP or Linphone, and a desk phone configured by hand, both need the same details: the SIP server, the port, and the extension’s username and password. These are the credentials of a SIP account, and anyone who has them can make calls on your bill, so treat them like a password and never paste them into an open email or a shared chat.

Many desk phones offer an easier route called auto-provisioning. You give the phone the address of a configuration server, or hand it out automatically through DHCP option 66; the phone identifies itself by its MAC address and downloads its account, clock, language and settings on its own.

1. Find the MAC address

   It is printed on the label under the phone and identifies that specific device.

2. Link the phone to an extension

   In the PBX, pair the brand and MAC address with the person who will use the phone.

3. Set the provisioning URL

   Enter the configuration server address in the phone’s web interface, or distribute it through DHCP.

4. Reboot and check

   On restart the phone downloads its configuration and registers. Confirm on the display or in the PBX that the extension shows as registered.

Headsets, microphones and the desk setup

With a webphone, the computer is the phone, so quality depends on what you plug into it. A laptop’s built-in microphone picks up the fan, the keyboard and the room, while speakers create echo that the caller hears as their own voice coming back. For anyone on the phone more than a few times a day, a headset is the first thing to sort out.

• Boom microphone headset: puts the mic close to the mouth and cuts background noise far better than earbuds with an in-line mic.
• USB or Bluetooth: USB headsets are stable and have no battery to charge; Bluetooth gives you freedom to move but some combinations drop to a lower-quality audio profile while the microphone is in use.
• Two ears or one: in an open-plan office a stereo headset blocks out more noise; at a front desk a single earpiece lets you hear people walking up.
• Default device: check in both the browser and the operating system which microphone and output are selected, especially after plugging in a new headset.
• Room: a quiet spot with soft surfaces does more for call quality than any noise filter.

Network and call quality: what to check

A VoIP call uses very little bandwidth, under 100 kbit/s each way even with uncompressed G.711, but it is sensitive to how steady the connection is. Three measures matter: latency, the time the voice takes to arrive; jitter, how much that time varies; and packet loss. As a reference point, ITU-T Recommendation G.114 suggests keeping one-way delay under 150 milliseconds for a natural conversation.

• Use a cable rather than Wi-Fi for fixed desks: busy Wi-Fi adds jitter and brief dropouts that sound like clipped words.
• Keep heavy uploads, such as backups or file syncing, from saturating the line during working hours, or give voice traffic priority with your router’s QoS settings.
• If IP phones register but audio only goes one way, check the router’s SIP ALG. It is often the culprit, and turning it off usually fixes it.
• On locked-down corporate networks, make sure the firewall allows traffic to the PBX servers; otherwise audio may fall back to TURN relays, if available, over a longer route.
• At home, a corporate VPN that sends all traffic through a distant office can add delay. Where you can, let voice go straight out to the internet.

The agent panel: phone, customer details and CRM in one place

The real win of a browser phone is not just skipping an install. It is having the call and the information on the same screen. When the phone rings, the agent sees who is calling, which queue the call came from, how long they waited and the contact’s details, instead of hunting for them in another window while the customer is talking.

In the Robottino agent panel the webphone stays registered in every section of the panel: agents set themselves available and take calls from their call queues through the computer headset. When a call ends they record the outcome and, if needed, schedule a callback. The panel also works inside an iframe on another site, such as your CRM, so the phone lives where agents already work. For a deeper data connection, see the guide on integrating your phone system with a CRM.

Anyone who prefers a softphone or desk phone can use it with the same extension. An agent is registered on one device at a time: whichever registers last, softphone, desk phone or webphone, replaces the previous one. Agents who work without the panel change their availability from the phone with *30 (available) and *31 (on break).

Common mistakes to avoid

Most early problems with browser calling come down to a handful of habits.

• Opening the webphone in several tabs or on several computers and losing track of which one rings. Keep the phone open in one place.
• Letting the computer go to sleep or closing the browser after a break: the webphone disconnects and calls stop arriving.
• Skipping browser updates: older versions can have trouble with audio and encryption.
• Using a laptop microphone in a noisy room and blaming the phone system.
• Sharing SIP credentials between colleagues: everyone needs their own extension, not least so reports and recordings are accurate.
• Forgetting to go on break: queues keep sending calls to an empty desk, which means unanswered rings. If this happens a lot, read how to reduce missed calls.

Frequently asked questions