Frequency-Adaptive Flow Matching for High-Fidelity Text-to-Speech Synthesis
Electronics · 10.11.2026 · 15 (22) · 5210
DOI: 10.3390/electronics15225210 ↗Kurz gefasst
Fachartikel in der Zeitschrift Electronics, online erschienen am 10.11.2026. Verfasst von Chi Liu und 4 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: China National Electric Apparatus Research Institute.
Abstract (Originalsprache)
Flow matching (FM) enables efficient acoustic modeling for text-to-speech (TTS), but fine spectral structures may remain imperfectly reconstructed. To address this issue, we propose frequency-adaptive flow matching (FAFM), which replaces the shared linear interpolation schedule with a frequency-dependent nonlinear schedule. By assigning different interpolation exponents to Mel-frequency bands, FAFM advances high-frequency components earlier along their conditional paths while preserving the endpoints of the linear baseline. The corresponding analytical target velocity modulates frequency-dependent evolution without introducing additional network modules. Evaluations across two TTS architectures and English and Chinese datasets show improvements in spectral reconstruction and several objective quality and intelligibility metrics. Results averaged over three independent training seeds, together with band-wise spectral analysis, support the effectiveness of moderate frequency modulation. Subjective listening tests show small numerical gains in perceived naturalness over standard CFM.
Förderung
- China National Electric Apparatus Research Institute
Zitieren
APA
Liu, C., Lu, P., Chen, H., Su, J., & Guo, H. (2026). Frequency-Adaptive Flow Matching for High-Fidelity Text-to-Speech Synthesis. Electronics, 15(22), 5210. https://doi.org/10.3390/electronics15225210
Vancouver
Liu C, Lu P, Chen H, Su J, Guo H. Frequency-Adaptive Flow Matching for High-Fidelity Text-to-Speech Synthesis. Electronics. 2026;15(22):5210. doi:10.3390/electronics15225210
BibTeX
@article{Liu2026,
author = {Liu, Chi and Lu, Pengwei and Chen, Hui and Su, Jiuheng and Guo, Haiyan},
title = {Frequency-Adaptive Flow Matching for High-Fidelity Text-to-Speech Synthesis},
journal = {Electronics},
year = {2026},
volume = {15},
number = {22},
pages = {5210},
doi = {10.3390/electronics15225210}
}
EUVIMED