Are reconstruction metrics still enough to eval neural audio codecs?
Gauri_the_great · x · 2026-11-19
With neural audio codecs now heavily used in generative speech models, the author questions how long evaluation can keep relying on reconstruction metrics. Most codecs still lean on rate-distortion objectives combining spectral reconstruction, adversarial losses, and quantization objectives — but these say little about the conditional distribution an AR model has to learn, and AR models dominate for speed.
More from Research
- MIT CSAIL paper: frozen LLM agents design their own self-improvement without being told how — omarsar0 · 2026-11-19
- Radiology benchmark RadLE is 100% human-verified; RadLE 2.1 results coming at RSNA — DrDatta_AIIMS · 2026-11-19
- Anthropic Says Claude Solved Two Open Problems in Black Hole Physics — ghumare64 · 2026-11-19
- Shu Ding's essay 'Metaphors': protocols, dimensional analysis, and Curry-Howard — shuding · 2026-11-19
- Why your fine-tuning job takes longer than the FLOPs say — Hugging Face Blog · 2026-11-19
- CyberArena benchmark: Lithos defense agent wins 3-0, blocking 12 of 12 attacks — JiaZhihao · 2026-11-19