{"ID":2861429,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2510.01903","arxiv_id":"2510.01903","title":"MelTok: 2D Tokenization for Single-Codebook Audio Compression","abstract":"Large Audio Language Models (LALMs) have emerged with strong performance across diverse audio understanding tasks and can be further enhanced by neural audio codecs. Transitioning from multi-layer residual vector quantizers to a single-layer quantizer has been shown to facilitate more efficient downstream language models decoding. However, the ability of a single codebook to capture fine-grained acoustic details remains limited, as the frequency-variant nature of 1D tokenizers leads to redundancy. To address this issue, we propose MelTok, a two-dimensional (2D) tokenizer that effectively compresses acoustic details of 44.1 KHz audio into a single codebook. The tokenizer encodes audio into a more compact representation than one-dimensional tokenizers. Furthermore, to recover audio from mel-spectrogram tokens, we propose a token-based vocoder. Both objective and subjective evaluations demonstrate that MelTok achieves quality comparable to multi-codebook codecs and outperforms existing state-of-the-art neural codecs with a single codebook on high-fidelity audio reconstruction. By preserving acoustic details, MelTok offers a strong representation for downstream understanding tasks.","short_abstract":"Large Audio Language Models (LALMs) have emerged with strong performance across diverse audio understanding tasks and can be further enhanced by neural audio codecs. Transitioning from multi-layer residual vector quantizers to a single-layer quantizer has been shown to facilitate more efficient downstream language mode...","url_abs":"https://arxiv.org/abs/2510.01903","url_pdf":"https://arxiv.org/pdf/2510.01903v3","authors":"[\"Jingyi Li\",\"Zhiyuan Zhao\",\"Zhisheng Zhang\",\"Yunfei Liu\",\"Lijian Lin\",\"Ye Zhu\",\"Jiahao Wu\",\"Qiuqiang Kong\",\"Yu Li\"]","published":"2025-10-02T11:17:37Z","proceeding":"cs.SD","tasks":"[\"cs.SD\",\"eess.AS\"]","methods":"[\"Language Model\"]","has_code":false}