Three VQ-based QAT techniques (cosine-similarity STE, hard-attention VQ, mixed VQ/LQ NAS) are proposed; they simplify training but do not consistently outperform existing methods.
”Soundstream: An end-to-end neural audio codec.” IEEE/ACM Transactions on Audio, Speech, and Language Processing 30 (2021): 495-507
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SE 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
RAT: RunAnyThing via Fully Automated Environment Configuration
Three VQ-based QAT techniques (cosine-similarity STE, hard-attention VQ, mixed VQ/LQ NAS) are proposed; they simplify training but do not consistently outperform existing methods.