[asterisk-dev] Asterisk 14.0.1 Now Available

Alexander Traud pabstraud at compuserve.com
Mon Oct 31 16:20:27 CDT 2016


>Has anybody updated the version of the [Opus transcoding] patch for 14 and/or master?

<https://github.com/traud/asterisk-opus>

The story continues there. I updated the code and I try my best to maintain that fork (because I use it myself). Currently, it should be compatible from 13.7 up to the latest 14 and Master version. If not, please, create an Issue Report or a Pull Request. Other contributions are welcome as well.

> it does take a small bits of maintenance


If there is anything I could do, to ease that, please do not hesitate.

> Opus is to become the new standard audio codec.

I know, you want to persuade Digium to give more attention to that audio codec and its features (Native PLC, Adaptive FEC, VAD/DTX/CNG). Yes, those features of a Media Gateway are important not only to Opus but other audio codecs as well, like 3GPP EVS. Actually, because of the complexity of modern  audio codecs (since the end of the nineties with the introduction of G.729), such features are a must-have. Without those features, distortion is a common issue with modern audio codecs.

Furthermore, I cannot restrain to comment on that statement: I am quite skeptical about the future of Opus. Currently, it is there because of WebRTC. Full stop! Mobile phones go for GSM, AMR, AMR-WB, and last years 3GPP EVS. Landline phones go for and continue to use G.711, G.726-32, G.722. If Opus gets lucky, the industry chooses Opus for multi-channel and music via landline. However last year, for music, the German company AVM went not for Opus but for its precursor CELT.


Finally, from my experiences with several implementations, Opus Codec seems to be quite challenging. Not many implementations leverage the parameter negotiation via the SDP attribute fmtp. It was a bit of work to add that to Asterisk, by the way. Therefore sometimes, tailoring of the Opus Codec is impossible and the data rate goes through the sky. When it comes to wide-band audio codecs, G.722 and AMR-WB might stay the winners because they are more limited.

Nevertheless, I would have nothing against a single audio codec for VoIP/SIP  - finally.




More information about the asterisk-dev mailing list