Saved in:
| Main Authors: | , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2504.20334 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910925914636288 |
|---|---|
| author | Liang, Yuzhe Liu, Wenzhe Qiang, Chunyu Niu, Zhikang Chen, Yushen Ma, Ziyang Chen, Wenxi Li, Nan Zhang, Chen Chen, Xie |
| author_facet | Liang, Yuzhe Liu, Wenzhe Qiang, Chunyu Niu, Zhikang Chen, Yushen Ma, Ziyang Chen, Wenxi Li, Nan Zhang, Chen Chen, Xie |
| contents | Flow matching has demonstrated strong generative capabilities and has become a core component in modern Text-to-Speech (TTS) systems. To ensure high-quality speech synthesis, Classifier-Free Guidance (CFG) is widely used during the inference of flow-matching-based TTS models. However, CFG incurs substantial computational cost as it requires two forward passes, which hinders its applicability in real-time scenarios. In this paper, we explore removing CFG from flow-matching-based TTS models to improve inference efficiency, while maintaining performance. Specifically, we reformulated the flow matching training target to directly approximate the CFG optimization trajectory. This training method eliminates the need for unconditional model evaluation and guided tuning during inference, effectively cutting the computational overhead in half. Furthermore, It can be seamlessly integrated with existing optimized sampling strategies. We validate our approach using the F5-TTS model on the LibriTTS dataset. Experimental results show that our method achieves a 9$\times$ inference speed-up compared to the baseline F5-TTS, while preserving comparable speech quality. We will release the code and models to support reproducibility and foster further research in this area. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_20334 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Towards Flow-Matching-based TTS without Classifier-Free Guidance Liang, Yuzhe Liu, Wenzhe Qiang, Chunyu Niu, Zhikang Chen, Yushen Ma, Ziyang Chen, Wenxi Li, Nan Zhang, Chen Chen, Xie Audio and Speech Processing Flow matching has demonstrated strong generative capabilities and has become a core component in modern Text-to-Speech (TTS) systems. To ensure high-quality speech synthesis, Classifier-Free Guidance (CFG) is widely used during the inference of flow-matching-based TTS models. However, CFG incurs substantial computational cost as it requires two forward passes, which hinders its applicability in real-time scenarios. In this paper, we explore removing CFG from flow-matching-based TTS models to improve inference efficiency, while maintaining performance. Specifically, we reformulated the flow matching training target to directly approximate the CFG optimization trajectory. This training method eliminates the need for unconditional model evaluation and guided tuning during inference, effectively cutting the computational overhead in half. Furthermore, It can be seamlessly integrated with existing optimized sampling strategies. We validate our approach using the F5-TTS model on the LibriTTS dataset. Experimental results show that our method achieves a 9$\times$ inference speed-up compared to the baseline F5-TTS, while preserving comparable speech quality. We will release the code and models to support reproducibility and foster further research in this area. |
| title | Towards Flow-Matching-based TTS without Classifier-Free Guidance |
| topic | Audio and Speech Processing |
| url | https://arxiv.org/abs/2504.20334 |