Recent state-of-the-art neural audio compression models have progressively adopted residual vector quantization (RVQ). Despite this success, these models employ a fixed number of codebooks per frame, which can be suboptimal in terms of rate-distortion tradeoff, particularly in scenarios with simple input audio, such as silence. To address this limitation, we propose variable bitrate RVQ (VRVQ) for audio codecs, which allows for more efficient coding by adapting the number of codebooks used per frame. Furthermore, we propose a gradient estimation method for the non-differentiable masking operation that transforms from the importance map to the binary importance mask, improving model training via a straight-through estimator. We demonstrate that the proposed training framework achieves superior results compared to the baseline method and shows further improvement when applied to the current state-of-the-art codec.
View on arXiv@article{chae2025_2410.06016,
title={ Variable Bitrate Residual Vector Quantization for Audio Coding },
author={ Yunkee Chae and Woosung Choi and Yuhta Takida and Junghyun Koo and Yukara Ikemiya and Zhi Zhong and Kin Wai Cheuk and Marco A. Martínez-Ramírez and Kyogu Lee and Wei-Hsiang Liao and Yuki Mitsufuji },
journal={arXiv preprint arXiv:2410.06016},
year={ 2025 }
}