F5-TTS (A Fairytaler that Fakes Fluent and Faithful Speech with Flow Matching) is a high-quality, zero-shot text-to-speech system from Shanghai Jiao Tong University's X-LANCE Lab, built on a Diffusion Transformer with ConvNeXt V2 architecture trained using flow matching. It supports natural speech synthesis with voice cloning from a short reference audio clip and features Sway Sampling—an inference-time flow step sampling strategy that greatly improves generation quality and speed. The v1 base model released in March 2025 delivers better training stability and inference performance, achieving state-of-the-art naturalness on multiple benchmarks.