Experiments show that TextPro-SLM achieves the lowest modality gap among leading SLMs at both 3B and 7B scales, while also delivering strong overall performance on paralinguistic understanding tasks, suggesting that reducing the modality gap from the input side is both effective and data-efficient.
Wen-Qian Cui, Xiao-Hui Li, Da-Xin Tan et al.· arXiv.org· 3 citations
NormToken is an iterative semantic token purification framework that alternates S2U and text-to-unit (T2U) training that demonstrates improved S2U--T2U agreement and parallel-utterance token consistency and reduced dependence on source duration information.
Han-Lin Zhang, Da-Xin Tan, De-Hua Tao et al.· 0 citations
Joint-Output On-Policy Distillation (JO-OPD) is proposed, which distills the model's stronger S2T policy into joint generation using student-generated S2TS trajectories, which reduces OMG across two interleaved generation architectures.
Da-Xin Tan, De-Hua Tao, Chen-Xi Deng et al.· 0 citations
Independently trained de-tokenizers show that the refined S2U and T2U tokens retain sufficient content for high-intelligibility voice conversion and text-to-speech synthesis and exhibit substantially improved cross-speaker consistency and reduced probe-recoverable speaker information.
Hanlin Zhang, Da-Xin Tan, De-Hua Tao et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
