F5-TTS
github.com
- Free plan
- macOS
- Linux
- Web
- API
- Self-hosted
What F5-TTS is
F5-TTS is an open-source text-to-speech project that generates speech from text using flow matching. Its design includes a Diffusion Transformer with ConvNeXt V2 and the E2 TTS Flat-UNet Transformer. The project offers command-line inference and a Gradio interface. The Gradio app supports chunk inference, multiple styles and speakers, and voice chat powered by Qwen2.5-3B-Instruct. Custom inference guidance covers adding language support. Users can install a Python package, run an editable local clone, or build with Docker. Training and fine-tuning work through Hugging Face Accelerate or a Gradio app. Installation guidance covers NVIDIA, AMD, and Intel GPUs plus Apple Silicon; AMD ROCm instructions specify Linux. Deployment guidance uses Triton and TensorRT-LLM. Requirements include Python 3.10 or newer and FFmpeg. Code uses the MIT License, while pretrained models use CC-BY-NC. Commercial use is listed as unavailable.
Who it suits
It suits developers and researchers who want self-hosted text-to-speech with command-line, API, or Gradio access. Users should note the stated noncommercial restriction for pretrained models and project use.
Good to know, and what to watch
- Open-source code under the MIT License
- Command-line and Gradio interfaces
- Supports multiple styles and speakers
- Training and fine-tuning workflows included
- Runs across several GPU and Apple setups
- Commercial use is listed as unavailable
- Pretrained models use CC-BY-NC
- Requires Python 3.10 or newer and FFmpeg
The verdict
F5-TTS provides several ways to run, customize, and deploy an open-source speech system. Its broad hardware guidance and interfaces help technical users, while licensing limits commercial use.
What the maker publishes
- API access
- Yes source
- Commercial use
- No source
- Custom languages
- The README links to custom inference instructions for adding language support. source
- Deployment
- The README describes a deployment solution using Triton and TensorRT-LLM. source
- Hardware and operating systems
- The installation instructions cover NVIDIA, AMD, and Intel GPUs and Apple Silicon; the AMD ROCm instructions specify Linux. source
- Hardware requirements
- The installation instructions cover NVIDIA, AMD, Intel, and Apple Silicon setups; the AMD ROCm instructions specify Linux. source
- Inference features
- The Gradio app supports basic text-to-speech with chunk inference, multi-style and multi-speaker generation, and voice chat powered by Qwen2.5-3B-Instruct. source
- Interfaces
- The project provides a Gradio web interface and command-line inference. source
- License
- The code is released under the MIT License, while the pre-trained models are licensed under CC-BY-NC. source
- Model availability
- The README links F5-TTS and E2 TTS base models on Hugging Face, Model Scope, and Wisemodel. source
- Model design
- The project describes F5-TTS as a Diffusion Transformer with ConvNeXt V2 and includes E2 TTS, a Flat-UNet Transformer. source
- Platforms
- web,linux,macos,api,self_hosted source
- Purpose
- F5-TTS is an open-source text-to-speech project that uses flow matching to generate fluent and faithful speech. source
- Related tools
- The project acknowledges Vocos and BigVGAN as vocoders and FunASR and faster-whisper among its evaluation tools. source
- Requirements
- The installation instructions call for Python 3.10 or newer and FFmpeg. source
- Runtime benchmark
- The README reports a 253 ms average latency for F5-TTS Base with Vocos at concurrency 2 on a single L20 GPU in its stated benchmark setup. source
- Support guidance
- The README recommends searching GitHub issues for help with problems encountered. source
- Training
- Training and fine-tuning are supported through Hugging Face Accelerate or a Gradio app. source
- Voice cloning
- Yes source
- Ways to run
- The project can be installed as a Python package, run from a local editable clone, or built and run with Docker. source
- Web interface
- The project provides Gradio commands for launching a web interface and a share link. source
Checked against the maker's pages on 8 Oct 2026.