Recent advancements in discrete token-based speech generation have highlighted the importance of efficient token-to-waveform synthesis in streaming and dialogue scenarios. Flow-matching acoustic decoders achieve high-quality token-to-mel generation, but their iterative sampling requires multiple neural function evaluations, limiting low-latency speech synthesis. MeanFlow reduces the sampling budget by modeling the average velocity over a temporal interval, yet maintaining high acoustic quality under extremely few-step token-to-mel generation remains challenging. To address this challenge, we propose X-Pred MeanFlow, a few-step streaming token-to-mel decoder that reparameterizes MeanFlow with mel-space prediction. The decoder predicts a generalized mel field and analytically derives the corresponding average velocity for sampling, thereby preserving the MeanFlow formulation while providing a direct acoustic prediction target. We further introduce layer-selective block-wise attention to enable continuous chunk-wise generation with bounded context. Experiments show that X-Pred MeanFlow improves few-step token-to-mel synthesis over Direct-$u$ MeanFlow and supports stable streaming generation. Speech samples are available.https://renxiaming.github.io/xpred-meanflow-stream-demo
Han-Ke Xie, Xia-Ming Ren, Qi-Rui Zhan et al.· 0 citations
We examined the longitudinal association between the C-reactive protein-triglyceride-glucose index (CTI) and cardiovascular disease (CVD) reported at the 2018 follow-up among middle-aged and older Chinese adults with arthritis, evaluated the dose–response pattern, and explored the potential statistical mediation by body mass index (BMI).This prospective longitudinal cohort study used the 2011 baseline and 2018 follow-up waves of the China Health and Retirement Longitudinal Study. We included 2113 participants aged 45 years or older with physician-diagnosed arthritis, no CVD at baseline, and complete data required for the primary analysis. Multivariable logistic regression was used to estimate odds ratios (ORs) for CVD reported at the 2018 wave. Restricted cubic splines assessed the dose–response pattern, and an exploratory mediation analysis evaluated BMI. At the 2018 follow-up, 467 participants (22.10%) reported CVD. In the fully adjusted model, each 1-unit increase in CTI was associated with greater odds of CVD (OR 1.223, 95% CI 1.009–1.484;
P
= 0.040). Compared with Q1, the OR for Q4 was 1.540 (95% CI 1.122–2.114;
P
= 0.008), and the quartile-median trend was significant (
P
for trend = 0.009), although Q2 was not statistically significant. The adjusted spline showed an overall association (
P
-overall = 0.034) without strong evidence of nonlinearity (
P
-nonlinear = 0.095). Exploratory mediation analysis estimated a mediated proportion of 36.75% (95% CI 7.62–217.64%), indicating substantial uncertainty. Higher baseline CTI was associated with greater odds of CVD reported at the 2018 follow-up among middle-aged and older Chinese adults with arthritis. The findings do not establish causality or clinical predictive utility, and the exploratory BMI mediation result should be interpreted cautiously.
Qiang Yuan, Ru-Hui Fu, Ning Zhang et al.· Scientific Reports· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
