Technology Category
Natural Language Processing: Sentiment Analysis, Stylistic Analysis, and Argument MiningMachine Learning: Deep Generative Models & AutoencodersComputer Vision: Generative Adversarial Networks (GANs) for Vision
Spotlight-TTS: Spotlighting the Style via Voiced-Aware Style Extraction and Style Direction Adjustment for Expressive Text-to-Speech
🤖 AI Summary
This work addresses two key challenges in expressive text-to-speech (TTS): insufficient style modeling and poor semantic alignment between style embeddings and speech decoding. To this end, we propose a voiced-aware style extraction mechanism and a style-direction adaptive adjustment module. Methodologically: (i) style features are extracted exclusively from voiced segments to suppress noise from silent regions; (ii) a temporally consistent style encoder—enforced via continuity constraints—and a learnable style-direction projection layer are introduced to enhance temporal coherence of style representations and improve alignment with decoder semantics. Integrated into a FastSpeech 2–based end-to-end framework, our approach achieves state-of-the-art performance on the Expressive TTS benchmark: MOS improves by 0.32, and style transfer accuracy increases by 12.6%. Open-sourced code and synthesized audio validate simultaneous gains in naturalness, style fidelity, and controllability.
Technology Category
Application Category
📝 Abstract
Recent advances in expressive text-to-speech (TTS) have introduced diverse methods based on style embedding extracted from reference speech. However, synthesizing high-quality expressive speech remains challenging. We propose Spotlight-TTS, which exclusively emphasizes style via voiced-aware style extraction and style direction adjustment. Voiced-aware style extraction focuses on voiced regions highly related to style while maintaining continuity across different speech regions to improve expressiveness. We adjust the direction of the extracted style for optimal integration into the TTS model, which improves speech quality. Experimental results demonstrate that Spotlight-TTS achieves superior performance compared to baseline models in terms of expressiveness, overall speech quality, and style transfer capability. Our audio samples are publicly available.
Problem
Research questions and friction points this paper is trying to address.
Extracting style from voiced speech regions for better expressiveness
Adjusting style direction for optimal TTS integration
Improving expressive speech synthesis quality and style transfer
Innovation
Methods, ideas, or system contributions that make the work stand out.
Voiced-aware style extraction for expressiveness
Style direction adjustment for integration
Superior performance in expressive TTS
🔎 Similar Papers
No similar papers found.
