Meta has announced work on an AI-powered audio compression method called “EnCodec,” that they state delivers ten times better compression than the MP3 format. The new method consists of a three-party system that is trained to compress audio to the desired target size. Meta researchers state that “…the key to lossy compression is to identify changes that will not be perceivable to humans, as perfect reconstruction is impossible at low bit-rates.”
As the technology develops, Meta aims to deliver high-quality audio across all network connections with smaller files. Researchers also say that EnCodec could eventually deliver “…rich metaverse experiences without requiring major bandwidth improvements.”
Click here to read more from Digital Music News.