Meta/Facebook has announced EnCodec, a new AI-powered audio codec, compressed to one-tenth the size of the MP3 file format. Meta says the technology can significantly improve the sound quality of speech at low bandwidths. Meta also published the paper “High Fidelity Neural Audio Compression” on the preprint platform arxiv. The new method consists of three parts. First, the encoder converts the uncompressed data into a latent space representation at a low frame rate; the quantizer quantizer then compresses the representation to a target size, while keeping track of the most important information for future reconstruction of the original. signal; the decoder finally converts the compressed data into audio in real time using a neural network on a single CPU. The researchers say they are the first to implement neural network techniques for compressing 48 kHz stereo.
This article is reprinted from: https://www.solidot.org/story?sid=73244
This site is for inclusion only, and the copyright belongs to the original author.