Opus is designed for both voice and music applications with minimum latency
The Internet Engineering Taskforce (IETF) has approved the Opus Interactive Audio Codec for standardisation. The final specifications for the audio compression format will most likely be released as an RFC in six to eight weeks, which is equivalent to recommending the codec as an internet standard.
The codec – developed by Jean-Marc Valin and Timothy B. Terriberry of Mozilla and Xiph.org, in collaboration with Koen Vos from Skype – is designed as a universal voice and music codec. Opus has very low algorithmic latency (starting at 5ms) which makes it ideal for interactive applications like voice over IP (VoIP) and music streaming. For VoIP use, the codec can use as little as 6kBit/s in a mono setup and the maximum bitrate for stereo music streams is 510kBit/s. The codec supports both constant and variable bitrates and is able to adjust bandwidth and sampling rates, as well as the size of transmitted audio frames, on the fly.
To reach the goal of being efficient for both voice and music encoding, Opus combines the linear prediction layer process which Koen Voes developed for Skype’s SILK codec and the constrained-energy lapped transform (CELT) algorithm developed by Valin and Terriberry. Vorbis developer Christopher Montgomery was also involved in creating CELT and both methods have also been submitted to the IETF for standardisation