As Webrtcworld's WebRTC Conference cranks up this week in Atlanta, the hype is already flooding my inbox. Embedding the open-source Opus voice codec into products seems to be the current WebRTC compatibility "path" for a number of companies, but it is a quick fix at best. WebRTC is more than just a codec or SIP end-point, to borrow a phrase from Jan Linden, and simply dropping in Opus into a software package, service, or phone will bring more issues to the table.
If you tell me you have a WebRTC "phone", I would assume, in the strictest sense, you have a programmable webpage running in a WebRTC browser that is built with HTML 5 and can be easily modified. If you have a SIP end-point capable of talking to a WebRTC "page" via Opus for voice calling and maybe supporting video through VP8, it's still just a SIP end-point that can talk to a WebRTC browser session.
Don't get me wrong -- it's nice that SIP-end points can support WebRTC without having to transcode. Alcatel-Lucent supports a WebRTC gateway (SBC would be a better term) that will transcode between SIP-end points and a WebRTC client without breaking a sweat; I saw them demo it at The Cable Show a couple of weeks ago, running a session between a Google Chrome browser running on a laptop and a Biscotti video conferencing gizmo, but the folks at Biscotti knew better than to say they suddenly had a WebRTC "phone".
A true WebRTC "phone" would be a device that natively runs a WebRTC-supporting browser and had the phone built in HTML as a webpage. You'd have whole bunch of capabilities other than a fixed phone handset supporting Opus, including all kinds of customization for vertical markets. Maybe the ultimate WebRTC "phone" is an Android tablet with a USB handset and/or numeric keypad plugged in, but I think the jury is out on how far and how soon we might reach that vision.
Finally, using Opus is going to open up some very hardcore audio hardware comparisons between vendors. Opus is a "superwideband" codec capable of supporting sampling rates of up to 48 kHz to deliver high-quality sound at up to 510 Kbps. The legacy narrowband PSTN samples at only 8 kHz or so, moving sound around at an optimum rate of maybe 56 Kbps analog in a 3 kHz range -- I don't care what sort of magic you claim to have, the minute you touch the PSTN, you go into narrowband and Opus doesn't do you any good.
Even in the HD voice arena, using G.722, there's a wide variation between vendor microphones, speaker quality, and acoustics engineering. Not all vendors that "support" G.722 are delivering a full HD voice experience because they haven't sweated the details on acoustic quality. On the other side of the curve, Polycom tends to engineer its pricey devices to support the best audio that can be delivered by the codec used in a voice or video call.
People will start running IP calls using Opus as the native codec and will notice differences in call quality between phones and hardware manufacturers – and that's before you even get to the pickup of background noise by single microphone phones used in speakerphone mode. Opus is a super-wideband codec, but it doesn't provide phones with magic powers to improve the sound quality -- that's a hardware issue.
Edited by
Blaise McNamee
