🎙️ F5-TTS FastAPI Wrapper

A lightweight FastAPI-based REST API that exposes the open-source F5-TTS voice synthesis model.
Generate speech, stream audio, and clone voices with simple API calls.

---

🚀 Features

• 🎧 Text-to-Speech (/tts, /tts-fast)
• 🔊 Real-time audio streaming (/tts-stream, /tts-stream-chunked)
• ⚙️ Configurable speed, steps, silence removal, and CFG
• 🧩 Easy to integrate in any web or SaaS app

---

🛠️ Setup

1️⃣ Clone the official F5-TTS repo

git clone https://github.com/SWivid/F5-TTS.git
cd F5-TTS

2️⃣ Add this FastAPI script

Copy or download app.py from this repo into the root of your F5-TTS folder.

3️⃣ Install dependencies

pip install -r requirements.txt

4️⃣ Run the API

uvicorn app:app --host 0.0.0.0 --port 8000

Then open in browser:
👉 http://localhost:8000/docs

---

🧪 Example Usage (Python)

import requests

url = "http://localhost:8000/tts"
files = {"ref_audio": open("ref.wav", "rb")}
data = {"gen_text": "Hello world, this is F5-TTS in action!"}

res = requests.post(url, files=files, data=data)
open("output.wav", "wb").write(res.content)

---

⚡ Example Usage (TypeScript / Node.js)

import fs from "fs";
import FormData from "form-data";
import axios from "axios";

const runTTS = async () => {
  const formData = new FormData();
  formData.append("ref_audio", fs.createReadStream("ref.wav"));
  formData.append("gen_text", "Hello world, this is F5-TTS from Node.js!");
  
  const response = await axios.post("http://localhost:8000/tts", formData, {
    headers: formData.getHeaders(),
    responseType: "arraybuffer",
  });

  fs.writeFileSync("output.wav", response.data);
  console.log("✅ Audio saved as output.wav");
};

runTTS().catch(console.error);

---

💡 Endpoints

Method	Endpoint	Description
GET	/health	Model/device info
POST	/tts	High-quality text-to-speech
POST	/tts-fast	Fast inference (lower quality)
POST	/tts-stream	Real-time WAV streaming
POST	/tts-stream-chunked	Chunked MP3 streaming
GET	/	Root status
GET	/mock-audio.mp3	Sample mock audio (for dev)

---

⚡ Notes

• Uses pretrained model: SWivid/F5-TTS/F5TTS_v1_Base
• Auto-detects GPU (CUDA) if available
• Output files saved to /generated folder

---

🧑‍💻 Author

Noman Nazir — Linkedin
AI-Powered Software & SaaS Solutions

---

📜 License

MIT License — Feel free to use, modify, and contribute!