Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2601.20362 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909020799893504 |
|---|---|
| author | Wang, Xiangbo Jiang, Wenbin Wang, Jin You, Yubo Fang, Sheng Wen, Fei |
| author_facet | Wang, Xiangbo Jiang, Wenbin Wang, Jin You, Yubo Fang, Sheng Wen, Fei |
| contents | Recent neural audio compression models often rely on residual vector quantization for high-fidelity coding, but using a fixed number of per-frame codebooks is suboptimal for the wide variability of audio content-especially for signals that are either very simple or highly complex. To address this limitation, we propose SwitchCodec, a neural audio codec based on Residual Experts Vector Quantization (REVQ). REVQ combines a shared quantizer with dynamically routed expert quantizers that are activated according to the input audio, decoupling bitrate from codebook capacity and improving compression efficiency. This design ensures full training and utilization of each quantizer. In addition, a variable-bitrate mechanism adjusts the number of active expert quantizers at inference, enabling multi-bitrate operation without retraining. Experiments demonstrate that SwitchCodec surpasses existing baselines on both objective metrics and subjective listening tests. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_20362 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Switchcodec: Adaptive residual-expert sparse quantization for high-fidelity neural audio coding Wang, Xiangbo Jiang, Wenbin Wang, Jin You, Yubo Fang, Sheng Wen, Fei Sound Artificial Intelligence Recent neural audio compression models often rely on residual vector quantization for high-fidelity coding, but using a fixed number of per-frame codebooks is suboptimal for the wide variability of audio content-especially for signals that are either very simple or highly complex. To address this limitation, we propose SwitchCodec, a neural audio codec based on Residual Experts Vector Quantization (REVQ). REVQ combines a shared quantizer with dynamically routed expert quantizers that are activated according to the input audio, decoupling bitrate from codebook capacity and improving compression efficiency. This design ensures full training and utilization of each quantizer. In addition, a variable-bitrate mechanism adjusts the number of active expert quantizers at inference, enabling multi-bitrate operation without retraining. Experiments demonstrate that SwitchCodec surpasses existing baselines on both objective metrics and subjective listening tests. |
| title | Switchcodec: Adaptive residual-expert sparse quantization for high-fidelity neural audio coding |
| topic | Sound Artificial Intelligence |
| url | https://arxiv.org/abs/2601.20362 |