Last year, Google announced the Lyra voice codec for low bit rates, which, combined with the open AV1 codec, enables voice chat over 56kbps Modem connections more than 20 years ago. Lyra can even work tenaciously at 3kbps after using machine learning and other techniques for very low bit rate voice compression. Google opened up Lyra’s source code last year, and today they’re announcing the availability of Lyra V2.
Google sums up Lyra V2 as “a better, faster, more versatile speech codec …… A new architecture that enjoys broader platform support offers scalable bit rate capabilities, has better performance, and produces higher quality audio.”
Lyra V2 leverages the SoundStream end-to-end neural audio codec and continues to show much better performance than the Opus audio codec, improved audio quality, and more.
The GitHub release notes mention that Lyra V2 can perform up to 5x faster on Android devices, with codec latency reduced from 100 ms to 20 ms, and now also supports Mac and Windows to complement Linux and Android support.