Google announced its new Lyra audio codec in February, which it would start using within Google Duo. The codec allows for higher quality audio at lower bandwidths, meaning it’s ideal for slower network speeds. On Friday, Google announced that it’s expanding this new codec to third-party developers, opening Lyra to a world of possibilities.
As part of our efforts to make the best codecs universally available, we are open sourcing Lyra, allowing other developers to power their communications apps and take Lyra in powerful new directions.
As a refresher, Lyra features an encoder to capture audio, which it compresses and then decodes using a generative model, a type of machine learning model. The model allows the audio to be recreated at particularly low bitrates, as low as 3kbps while retaining comparatively high-quality output. You can get a more detailed explanation on Google’s AI blog, but Lyra is essentially ideal for use on the best cheap Android phones or slow networks. Google highlights the importance of technology like Lyra, particularly in emerging markets that don’t have fast internet speeds in a time when telecommuting has become a norm.
Making Lyra open source means that the codec can be used in more communication applications, but Google hopes developers will come up with new ways to use the codec. That could include uses for non-speech applications or even music, which would be ideal for platforms like Spotify that are continuously expanding to more global markets, but the possibilities are endless.
Google says that Lyra is written in C++ “for speed, efficiency, and interoperability” and that today’s beta release and demo on GitHub enables developers to provide feedback quickly.