MOSS-Audio-Tokenizer: Scaling Audio Tokenizers for Future Audio Foundation Models
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866918333329178624 |
|---|---|
| author | Gong, Yitian Chen, Kuangwei Fei, Zhaoye Yang, Xiaogui Chen, Ke Wang, Yang Huang, Kexin Chen, Mingshu Li, Ruixiao Cheng, Qingyuan Li, Shimin Qiu, Xipeng |
| author_facet | Gong, Yitian Chen, Kuangwei Fei, Zhaoye Yang, Xiaogui Chen, Ke Wang, Yang Huang, Kexin Chen, Mingshu Li, Ruixiao Cheng, Qingyuan Li, Shimin Qiu, Xipeng |
| contents | Discrete audio tokenizers are fundamental to empowering large language models with native audio processing and generation capabilities. Despite recent progress, existing approaches often rely on pretrained encoders, semantic distillation, or heterogeneous CNN-based architectures. These designs introduce fixed inductive biases that limit reconstruction fidelity and hinder effective scaling. In this paper, we argue that discrete audio tokenization should be learned fully end-to-end using a homogeneous and scalable architecture. To this end, we first propose CAT (Causal Audio Tokenizer with Transformer), a purely Transformer-based architecture that jointly optimizes the encoder, quantizer, and decoder from scratch for high-fidelity reconstruction. Building on the CAT architecture, we develop MOSS-Audio-Tokenizer, a large-scale audio tokenizer featuring 1.6 billion parameters, pre-trained on 3 million hours of diverse, general audio data. We show that this simple, fully end-to-end approach built from homogeneous, causal Transformer blocks scales gracefully and supports high-fidelity reconstruction across diverse audio domains. Across speech, sound, and music, MOSS-Audio-Tokenizer consistently outperforms prior codecs over a wide range of bitrates, while exhibiting predictable improvements with increased scale. Notably, leveraging the discrete tokens from our model, we develop the first purely autoregressive TTS model that surpasses prior non-autoregressive and cascaded systems. Furthermore, MOSS-Audio-Tokenizer enables competitive ASR performance without auxiliary encoders. Our findings position the CAT architecture as a unified, scalable interface for the next generation of native audio foundation models. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_10934 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | MOSS-Audio-Tokenizer: Scaling Audio Tokenizers for Future Audio Foundation Models Gong, Yitian Chen, Kuangwei Fei, Zhaoye Yang, Xiaogui Chen, Ke Wang, Yang Huang, Kexin Chen, Mingshu Li, Ruixiao Cheng, Qingyuan Li, Shimin Qiu, Xipeng Sound Audio and Speech Processing Discrete audio tokenizers are fundamental to empowering large language models with native audio processing and generation capabilities. Despite recent progress, existing approaches often rely on pretrained encoders, semantic distillation, or heterogeneous CNN-based architectures. These designs introduce fixed inductive biases that limit reconstruction fidelity and hinder effective scaling. In this paper, we argue that discrete audio tokenization should be learned fully end-to-end using a homogeneous and scalable architecture. To this end, we first propose CAT (Causal Audio Tokenizer with Transformer), a purely Transformer-based architecture that jointly optimizes the encoder, quantizer, and decoder from scratch for high-fidelity reconstruction. Building on the CAT architecture, we develop MOSS-Audio-Tokenizer, a large-scale audio tokenizer featuring 1.6 billion parameters, pre-trained on 3 million hours of diverse, general audio data. We show that this simple, fully end-to-end approach built from homogeneous, causal Transformer blocks scales gracefully and supports high-fidelity reconstruction across diverse audio domains. Across speech, sound, and music, MOSS-Audio-Tokenizer consistently outperforms prior codecs over a wide range of bitrates, while exhibiting predictable improvements with increased scale. Notably, leveraging the discrete tokens from our model, we develop the first purely autoregressive TTS model that surpasses prior non-autoregressive and cascaded systems. Furthermore, MOSS-Audio-Tokenizer enables competitive ASR performance without auxiliary encoders. Our findings position the CAT architecture as a unified, scalable interface for the next generation of native audio foundation models. |
| title | MOSS-Audio-Tokenizer: Scaling Audio Tokenizers for Future Audio Foundation Models |
| topic | Sound Audio and Speech Processing |
| url | https://arxiv.org/abs/2602.10934 |