What Is a SIP Softphone? A Practical Guide

A SIP softphone is a software application that lets a computer, phone or tablet make and receive voice calls over an IP network using the Session Initiation Protocol (SIP). It replaces the hardware desk phone with an app. The app registers to a SIP server with an account, uses SIP messages to start and end calls, and sends the audio as RTP packets. If the server is connected to the public telephone network, the softphone can call ordinary phone numbers too.

What does "softphone" mean?

The word combines "software" and "telephone". A hardphone is a physical device with a handset, keypad and screen. A softphone provides the same functions in software and uses the host device's microphone, speaker and display. The user dials, answers, holds and transfers calls from a window or a mobile screen instead of a handset.

A SIP softphone is a softphone that speaks SIP, the open signalling standard defined in RFC 3261. Because SIP is a standard, a SIP softphone from one vendor can register to a PBX or hosted service from another vendor. This is the main difference from closed calling apps, which only talk to their own service.

How does a SIP softphone work?

Three protocols do most of the work.

1. SIP handles signalling. It registers the softphone with the server, invites the other party to a call, negotiates the call details, and ends the call. See What is SIP? for the message flow.
2. SDP describes the media. Inside SIP messages, a small text block defined in RFC 8866 lists the IP address, port and codecs each side can use.
3. RTP carries the audio. Once both sides agree, the microphone audio is encoded with a codec and sent as a stream of UDP packets as defined in RFC 3550.

A typical call looks like this. The softphone starts and sends a REGISTER request to the server so the server knows where to reach it. When the user dials, the softphone sends an INVITE containing an SDP offer. The server routes the INVITE to the destination. When the far end answers, a 200 OK carries the SDP answer. Both sides now know where to send audio and begin exchanging RTP. When someone hangs up, a BYE ends the session.

What do you need to use one?

• A SIP account. This comes from a PBX such as Asterisk, FreePBX or Issabel, or from a VoIP provider. It includes a server address, a username, a password and sometimes a separate authentication ID. See What is a SIP account?.
• A network connection. Wi-Fi, Ethernet or mobile data. Voice needs low delay and low packet loss more than high bandwidth. A single G.711 call uses roughly 80 to 90 kbit/s in each direction including packet overhead.
• Audio hardware. The built-in microphone and speaker work. A headset avoids echo and background noise.
• Reachable ports. SIP commonly uses port 5060 for UDP and TCP and 5061 for TLS. RTP uses a range of UDP ports chosen by the server and the client. Firewalls and NAT routers sometimes interfere. See Common SIP ports.

What features do SIP softphones usually offer?

Most SIP softphones provide the basics: dial, answer, hang up, mute, hold and resume, a keypad for DTMF tones, and a call history. Many add call transfer, several simultaneous calls, multiple accounts, and call recording. Security features include TLS for signalling and SRTP for media.

Features vary by product. Some softphones support video, instant messaging and presence. Others focus on voice only. Some are built for a specific provider and are locked to it; others are generic and accept any SIP account.

SIPdia, for example, is a voice-only SIP softphone for macOS, iPhone and iPad. It supports UDP, TCP and TLS transports, SRTP media encryption, STUN and ICE for NAT traversal, multiple accounts, hold, transfer and local call recording. More than one account and call recording are part of SIPdia Premium. It uses the Opus and G.711 codecs and sends DTMF as RTP events. It does not offer video, chat or presence.

Softphone or desk phone: which should you choose?

A desk phone stays on and works independently of a computer. It is a good fit for a reception desk or a shared space. A softphone is a good fit for people who already work at a computer or on a mobile device, for remote work, and for anyone who wants to carry their extension with them. Many organisations use both and register the same extension on a desk phone and a softphone at once, so either device rings.

A softphone only receives calls while it is running and registered. On a laptop that sleeps, or on a phone where the app is in the background, incoming calls may be missed unless the app has a push notification mechanism. Check how a given softphone behaves before relying on it for inbound calls.

How does a SIP softphone handle security?

SIP signalling is plain text by default. Anyone on the path can read who is calling whom, and the authentication exchange uses a digest challenge that protects the password but not the rest of the message. Transport Layer Security (TLS) wraps the SIP connection in encryption. Secure RTP (SRTP) encrypts the audio. A softphone that supports both can keep the signalling and the audio private between the device and its server, even on untrusted networks. See SIP over UDP, TCP and TLS and What is SRTP?.

Security also depends on the server. A PBX exposed directly to the internet on port 5060 attracts automated attacks. Good practice is to restrict access with a VPN or IP allow-list, use strong passwords, and enable rate limiting on the server. How to protect a PBX from SIP scanning covers the server side.

What problems are common with softphones?

• Registration fails. Usually a wrong server address, port, transport or password. See SIP registration explained.
• One-way or no audio. Usually a NAT or firewall problem. The signalling gets through but RTP does not. STUN, ICE and the correct NAT settings on the server fix most cases. See How to fix one-way audio in VoIP calls.
• Choppy audio. Packet loss or jitter on the network. Wi-Fi congestion and overloaded uplinks are common causes. A codec like Opus tolerates loss better than G.711.
• Calls drop after a fixed time. A drop after about 30 seconds usually means the ACK or the 200 OK never arrived, often because of NAT or a router's SIP ALG. Drops at a longer fixed interval point to a session timer mismatch or a NAT mapping that expires. Keepalives and correct NAT settings on the server help.

Next steps

Browse the SIPdia features to see what a modern SIP softphone offers on macOS and iPhone. To understand the protocol behind every call, read What is SIP and how does SIP calling work?. To compare softphones with hardware, see SIP client vs VoIP phone.