Last week, Meta announced a new digital audio format called EnCodec, which can produce compressed audio at 64kbps without losing quality. According to the company’s report, the format is ten times more efficient than MP3 and, aside from digital music, may have its most dramatic impact on the quality of phone conversations starting from places with poor bandwidth and lower coverage.
The scientific article describing the procedure was published October 25 by Alexander DeVos, Jade Kobet, Gabriel Senev, and Yossi Ade, who developed EnCodec.
Compression consists of three steps. In the first step, the audio is compressed using the already known method, similar to MP3, in the next step, the file is compressed to the required size, while artificial intelligence collects the data required for recovery, also contained in the .
In the final step, a neural algorithm running on a processor restores the sound during playback. Of course, this is not the first sound compression technology to use neural networks.
Of course, Meta highlighted that “excess pressure” is excellent for getting excellent metaverseExperience without overburdening bandwidth – but if the metaverse became fashionableWhich can be missed by chance, can of course be of great help in lightening the load of networks burdened with broadcast media.
