AMR vs Opus & AAC

The AMR-WB (AMR Wideband) codec is the de facto standard for HD voice in the cellular arena. It has been turned up in over 50 3G networks (and growing) and is the standard codec for Voice over LTE (VoLTE). However, onerous licensing issues for software developers are limiting its proliferation beyond cell phones and mobile networks, opening the way for the open-source Opus codec and Fraunhofer's "Full HD voice" AAC codec.

Patents for AMR-WB are held by Ericsson, France Telecom/Orange Nokia and VoiceAge. A patent pool  -- basically, a one-stop place to license the patents  -- for AMR-WB device usage was announced in 2009. Managed by VoiceAge, the licensing is designed for application to consumer devices -- basically phones and tablets.

Software developers, on the other hand, have to negotiate with each of the individual patent holders. I've heard from two companies -- one that makes soft clients, the other core network server functions -- who went through the process. Both told the same story: Each patent holder has a different set of terms, conditions, and pricing. That's four different contracts, unlike the one-stop patent pool VoiceAge runs for device manufacturers.

Opus has become all the rage among the developer community and early adopters, driven by its incorporation into the browser for WebRTC. It's open source, royalty free, you use it, and it supports everything from narrowband voice in limited bandwidth conditions to streaming music at 510 kbps with sampling rates between 8 kHz to 48 kHz. Since you can find WebRTC in Google's Chrome and Mozilla's Firefox browser, you can get a high-quality audio experience by just pulling up a Web page.

Among the services and companies that have vowed support for Opus are CounterPath and TruPhone. CounterPath is the largest commercial soft client shop around while TruPhone will join Line2 in adding Opus support for its service/apps when it comes to client-to-client communications.

Compare AMR-WB to Opus or the royalty-enabled Fraunhofer AAC codecs and it becomes a grim contest. Opus and AAC both provide a wider and more flexible range of uses, with either simplifying implementation of sound playback to a single codec set capable of delivering everything from narrowband voice to high quality streaming music. Opus is royalty-free while you typically end up getting AAC thrown into/onto mobile devices via versions of Android and Apple iOS.

AMR-WB has two attributes going for it at this time: Bandwidth and CPU usage/battery life. A full blown AMR-WB call uses 24 kbps. Opus uses less bandwidth in the wideband mode, but has the potential of running up to 128 kbps for so-called "full band" stereo usage. Service providers would no doubt like to have the option to throttle bandwidth usage in certain use cases.

Battery life becomes the most interesting question. Since Opus is a software-based codec, processing takes place using RAM and CPU time, thereby placing more power demands on the battery.   But since nobody has (yet) run a side-by-side power comparison between an Opus call of one minute and 10 minutes to an AMR-WB call, it isn't clear how much of a power hit takes place. AMR-WB advocates would argue that on a performance-per-watt basis, its codec would beat Opus and AAC. In the real world, with plenty of people already opting to use Skype and Apple Facetime videoconferencing, the arguments for AMR-WB power efficiency might be academic rather than user driven. Apply Moore's Law for continued performance-per-watt efficiency and potentially some silicon tinkering/optimization for Opus usage and the argument continues to diminish.


references:
http://www.webrtcworld.com/topics/webrtc-world/articles/339560-amr-wb-its-future-hd-voice.htm