Just this week I experienced high-definition voice calling for the first time through my cell carrier. I knew it was HD voice because the call quality was noticeably better — and because I cover the VoIP industry, so I knew my carrier was taking advantage of voice-over-LTE (News - Alert) (VoLTE) technology.
Voice-over-IP (VoIP) and its variants such as VoLTE are more than just a fringe way to make calls these days. With carriers starting to run their networking on VoIP behind the scenes, and consumers embracing over-the-top apps such as Skype (News - Alert), WhatsApp and Line for their calling, VoIP is fast becoming the way that calling is made.
Many people don’t completely understand VoIP, however. VoIP is the general term for a family of methodologies, communication protocols and transmission technologies that deliver voice and multimedia sessions over IP networks. The steps involved in starting a VoIP call are signaling and media channel setup, digitization of the analog voice signal, compression, packetization, and transmission as IP packets over a packet-switched network. The same process is repeated on the receiving side.
Just as there are several formats for video streaming over the Internet, so too are there several codecs for VoIP. Some implementations use a narrow band (G729 32 kbps) and compressed speech, while others support high definition codecs such as G722 64 kbps. Almost all IP phones are G722 HD voice-ready, which gives the kind of call quality I experienced recently with my cell carrier.
While most Internet traffic uses TCP/IP for transmission, this is not sufficient for VoIP since TCP/IP has error correction and that is a problem for real-time applications that can’t wait for the error to be corrected; with VoIP, it is better to ignore the missed packet of data and keep going with the transmission.
Enabling this is user datagram protocol (UDP (News - Alert)), which has no handshaking or error correction and therefore allows for continuous packet streaming in real time protocol (RTP).
Of course, the problem with lacking error correction is packet loss. To combat this, VoIP uses very small packets of data so if any is lost it is hardly noticeable.
The cost for small packets is plenty of overhead, though, as there is more header information. A call using the G.729 codec will require 50 packets of 20 bytes worth of audio data for one second of audio. But due to all the header information, each of these 50 packets will actually be 60 bytes in size.
A lot goes into a VoIP call.
Edited by Rory J. Thompson