Salvato in:
Dettagli Bibliografici
Autori principali: Li, Bohan, Lian, Shi, Wang, Hankun, Guo, Yiwei, Xi, Yu, Li, Zhihan, Zheng, Da, Zhang, Colin, Yu, Kai
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:https://arxiv.org/abs/2605.29948
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916071829667840
author Li, Bohan
Lian, Shi
Wang, Hankun
Guo, Yiwei
Xi, Yu
Li, Zhihan
Zheng, Da
Zhang, Colin
Yu, Kai
author_facet Li, Bohan
Lian, Shi
Wang, Hankun
Guo, Yiwei
Xi, Yu
Li, Zhihan
Zheng, Da
Zhang, Colin
Yu, Kai
contents Unified speech foundation models require a holistic tokenization space that is both learnable by language models and decodable into high-quality waveforms. Existing speech tokenizers, however, often fail to satisfy these requirements simultaneously, leading to increased architectural complexity and more involved training designs. We propose HoliTok, a continuous Holistic speech Tokenization model designed for unified generation-understanding modeling. HoliTok encodes 48~kHz speech into a compact 25~Hz sequence of 128-dimensional latents. It is trained with a progressive strategy that jointly preserves signal-level fidelity, incorporates semantic information, and maintains strong latent learnability. Based on this tokenization, we build a unified AR+DiT model for speech synthesis and recognition, where the same latent sequence supports both generation-specific and unified generation-understanding tasks. Experiments show that HoliTok achieves competitive reconstruction fidelity, improves generative learnability for high-quality and controllable synthesis, and, among the evaluated representations, is the only one that operates robustly in our unified generation-understanding architecture without additional optimization tricks. These results suggest that HoliTok serves as an effective speech tokenizer and a foundational representation interface for unified spoken language modeling. The code is available at: https://github.com/bovod-sjtu/HoliTok.
format Preprint
id arxiv_https___arxiv_org_abs_2605_29948
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle HoliTok:A Coutinuous Holistic Tokenization with Robust Dual Capabilities of Speech Generation and Understanding
Li, Bohan
Lian, Shi
Wang, Hankun
Guo, Yiwei
Xi, Yu
Li, Zhihan
Zheng, Da
Zhang, Colin
Yu, Kai
Sound
Artificial Intelligence
Audio and Speech Processing
Unified speech foundation models require a holistic tokenization space that is both learnable by language models and decodable into high-quality waveforms. Existing speech tokenizers, however, often fail to satisfy these requirements simultaneously, leading to increased architectural complexity and more involved training designs. We propose HoliTok, a continuous Holistic speech Tokenization model designed for unified generation-understanding modeling. HoliTok encodes 48~kHz speech into a compact 25~Hz sequence of 128-dimensional latents. It is trained with a progressive strategy that jointly preserves signal-level fidelity, incorporates semantic information, and maintains strong latent learnability. Based on this tokenization, we build a unified AR+DiT model for speech synthesis and recognition, where the same latent sequence supports both generation-specific and unified generation-understanding tasks. Experiments show that HoliTok achieves competitive reconstruction fidelity, improves generative learnability for high-quality and controllable synthesis, and, among the evaluated representations, is the only one that operates robustly in our unified generation-understanding architecture without additional optimization tricks. These results suggest that HoliTok serves as an effective speech tokenizer and a foundational representation interface for unified spoken language modeling. The code is available at: https://github.com/bovod-sjtu/HoliTok.
title HoliTok:A Coutinuous Holistic Tokenization with Robust Dual Capabilities of Speech Generation and Understanding
topic Sound
Artificial Intelligence
Audio and Speech Processing
url https://arxiv.org/abs/2605.29948