반응형
[Paper 리뷰] SAC: Neural Speech Codec with Semantic-Acoustic Dual-Stream Quantization
SAC: Neural Speech Codec with Semantic-Acoustic Dual-Stream QuantizationSpeech codec은 semantically-rich representation과 high-quality reconstruction에 대한 trade-off가 존재함SACSemantic-acoustic dual-stream quantization을 활용해 semantic, acoustic modeling을 disentangling두 개의 dedicated stream을 각각의 respective role에 맞게 optimize논문 (ACL 2026) : Paper Link1. IntroductionSpeech tokenizer는 continuous speech wavefor..
Paper/Neural Codec
2026. 7. 1. 13:00
반응형