“The GANs are backdooring their way into really interesting outcomes, though. They're fantastic for compression: You compress the hell out of an input image or audio, then use the compressed features as conditioning for the GAN. This works great for super-resolution on images and speech compression. eg, for speech compression: https://arxiv.org/abs/2107.0…”
SoundStream: An End-to-End Neural Audio Codec
We present SoundStream, a novel neural audio codec that can efficiently compress speech, music and general audio at bitrates normally targeted by speech-tailored codecs. SoundStream relies on a model architecture composed by a fully convolutional encoder/decoder network and a residual vector quantizer, which are trained jointly end-to-end. Training leverages recent advances in text-to-speech and speech enhancement, which combine adversarial and reconstruction losses to allow the generation of high-quality audio content from quantized embeddings. By training with structured dropout applied to quantizer layers, a single model can operate across variable bitrates from 3kbps to 18kbps, with a negligible quality loss when compared with models trained at fixed bitrates. In addition, the model is amenable to a low latency implementation, which supports streamable inference and runs in real time on a smartphone CPU. In subjective evaluations using audio at 24kHz sampling rate, SoundStream at 3kbps outperforms Opus at 12kbps and approaches EVS at 9.6kbps. Moreover, we are able to perform joint compression and enhancement either at the encoder or at the decoder side with no additional latency, which we demonstrate through background noise suppression for speech.
候选层:至少 2 个不同的人、在 2 个不同月份提到过,还没到主书架的 3 人 3 个月。没有做立场标注,所以只写“被提及”。
HN 书评
推荐在前,同类按时间 · 每篇最多保留 20 条,统计用全部提及“The author, Fabrice Bellard, is well-known for his work on QEMU (among other things), but I'd treat the rest of the posts as "idea sharing" at an early stage. For example, while BPG (Better Portable Graphics) didn't take off, the core idea (image formats based on video codecs) can be found in WebP (VP8, VP9) and AVIF (AV1) and HEIF (HEVC/H.265). For this…”
“VAE objectives are useful for pushing embeddings into a Gaussian distribution. Here's some work on low-latency neural compression that you might find interesting: https://arxiv.org/abs/2107.03312”