BAMU：面向冻结预训练神经语音编解码器的比特流感知边际效用分配
BAMU: Bitstream-Aware Marginal-Utility Allocation for Frozen Pretrained Neural Speech Codecs
浏览论文内容
中文总结 AI 辅助
针对预训练神经语音编解码器固定RVQ深度忽略量化难度时间变化的问题，提出BAMU框架，通过预测器与分配器优化RVQ深度分配，在EnCodec和DAC上实现MOS提升。
中文摘要 AI 辅助
预训练神经语音编解码器通常对所有帧使用固定的残差向量量化（RVQ）深度，忽略了量化难度的时间变化。我们提出BAMU，一种面向冻结预训练编解码器的比特流感知动态RVQ分配框架。一个轻量、与速率无关的预测器估计帧级和层级的边际潜在失真减少量，而一个受约束的分配器在精确的序列化大小预算下选择前缀有效深度。在LibriSpeech数据集上对EnCodec和DAC进行的实验，结合VCTK评估，显示EnCodec持续获得增益，DAC主要在中高码率下得到改进。一项含30名听者的研究证实，在匹配的固定深度编码上，平均意见得分（MOS）从3.449提升至3.780。
英文摘要
Pretrained neural speech codecs typically use a fixed residual vector quantization (RVQ) depth for all frames, ignoring temporal variation in quantization difficulty. We propose BAMU, a bitstream-aware dynamic RVQ allocation framework for frozen pretrained codecs. A lightweight, rate-independent predictor estimates frame- and layer-wise marginal latent-distortion reductions, while a constrained allocator selects prefix-valid depths under an exact serialized-size budget. Experiments on EnCodec and DAC over LibriSpeech, together with VCTK evaluation, show consistent EnCodec gains and DAC improvements mainly at medium and high rates. A 30-listener study confirms a MOS improvement from 3.449 to 3.780 over matched fixed-depth coding.
