English

HH-Codec: High Compression High-fidelity Discrete Neural Codec for Spoken Language Modeling

Sound 2025-07-28 v1 Artificial Intelligence Audio and Speech Processing

Abstract

Discrete speech tokenization is a fundamental component in speech codecs. However, in large-scale speech-to-speech systems, the complexity of parallel streams from multiple quantizers and the computational cost of high-time-dimensional codecs pose significant challenges. In this paper, we introduce HH-Codec, a neural codec that achieves extreme compression at 24 tokens per second for 24 kHz audio while relying on single-quantizer inference. Our approach involves a carefully designed Vector Quantization space for Spoken Language Modeling, optimizing compression efficiency while minimizing information loss. Building on this, we propose an asymmetric encoder-decoder architecture (Audio-VQ-Mel-Audio) that leverages dual supervision and progressive training to enhance reconstruction stability and fidelity. HH-Codec achieves state-of-the-art performance in speech reconstruction with an ultra-low bandwidth of 0.3 kbps. We further evaluate its effectiveness in codebook utilization and generative model adaptation, with extensive ablations validating the necessity of each module. HH-Codec is available at https://github.com/opendilab/HH-Codec.

Keywords

Cite

@article{arxiv.2507.18897,
  title  = {HH-Codec: High Compression High-fidelity Discrete Neural Codec for Spoken Language Modeling},
  author = {Rongkun Xue and Yazhe Niu and Shuai Hu and Zixin Yin and Yongqiang Yao and Jing Yang},
  journal= {arXiv preprint arXiv:2507.18897},
  year   = {2025}
}
R2 v1 2026-07-01T04:18:06.089Z