# HearItServer: Your Offline TTS Server for Local Speech Synthesis

Nowadays AI-driven text-to-speech (TTS) solutions are dominated by cloud-based APIs, **HearItServer** emerges as a powerful alternative, bringing blazing-fast speech synthesis to local machines. Built on top of [**Kokoro-ONNX**](https://github.com/thewh1teagle/kokoro-onnx), the fastest and most efficient open-source TTS model, **HearItServer** provides developers with a ready-to-use, high-performance text-to-speech solution that can seamlessly integrate into their applications, enabling offline speech synthesis without requiring an internet connection.

I built HearItServer as a **core component** of a larger project I'm working on at the moment, a tool designed to help users read books, documents, and other text-based content **faster and more efficiently**. My goal is to **develop an app that enables users to consume more books** while making reading more engaging, all offline. HearItServer powers the **offline TTS** functionality of this project, but I realized it could also be useful to **developers looking for a lightweight, private, and fast text-to-speech solution**. So, I decided to make it **free and open** for others to build on.

If you need **real-time speech synthesis** without latency, data privacy concerns, or API rate limits, this is the **ultimate local TTS solution**.

---

## **Why Use HearItServer?**

Unlike traditional TTS services that require online APIs, HearItServer is designed to run **entirely on your local machine**. This means:

✅ **Lightning-Fast Inference** – Thanks to **Kokoro-ONNX**, the inference is optimized for speed.

✅ **Privacy-Preserving** – No data is sent to external servers, making it ideal for secure environments.

✅ **Fully Offline** – No need for API keys or internet connectivity.

✅ **Easy Integration into any application** – Exposes a simple **REST API** for seamless integration into any application you built.

---

## **How It Works**

HearItServer is essentially a lightweight Flask-based REST API that hosts **Kokoro-ONNX**, allowing any application to send text and receive **high-quality, natural-sounding speech** in response. This makes it **incredibly easy to integrate** into desktop applications, automation workflows, and AI assistants.

---

## **Setting Up HearItServer**

### **1️⃣ Install HearIt**

Download and install the **HearItServer** application on your machine. Once installed, launch it, and a **menu bar icon** will appear on macOS.

![System menu showing options for HearItServer: "Start TTS Server," "Stop TTS Server," "Browse docs," and "Quit."](https://cdn.hashnode.com/res/hashnode/image/upload/v1737312421495/8fd6bd6e-0d68-462c-99dd-45ef75c03dab.png align="center")

### **2️⃣ Start the TTS Server**

Click on the menu icon and select **"Start TTS Server"**. The server will now be running locally at:

```python
http://localhost:7008
```

---

## **Using the API (100% local)**

The HearItServer provides a simple API endpoint to generate speech from text.

### **Endpoint:**

```python
POST http://localhost:7008/v1/audio/speech
```

### **Request Body (JSON):**

```python
{
  "text": "Hello, this is a test message!",
  "voice": "af_sarah",
  "speed": 1.0,
  "lang": "en-us"
}
```

### **Available Voices:**

* `af_sarah`
    
* `af_bella`
    
* `af_nicole`
    
* `af_sky`
    
* `am_adam`
    
* `am_michael`
    
* `bf_emma`
    
* `bf_isabella`
    
* `bm_george`
    
* `bm_lewis`
    

### **Response:**

* **Success**: A `.wav` file is returned as a binary response.
    
* **Error**: A JSON object containing an error message.
    

---

## **Example: Using HearItServer in TypeScript**

To integrate HearIt into your application, you can send requests using TypeScript and **Axios**:

```python
import axios from 'axios';
import * as fs from 'fs';

const url = "http://localhost:7008/v1/audio/speech";
const headers = { "Content-Type": "application/json" };
const data = {
    text: "Hello, world!",
    voice: "af_sarah",
    speed: 1.0,
    lang: "en-us"
};

axios.post(url, data, { responseType: 'arraybuffer' })
    .then(response => {
        fs.writeFileSync("output.wav", Buffer.from(response.data));
        console.log("Audio saved as output.wav");
    })
    .catch(error => {
        console.error("Error:", error.response ? error.response.data : error.message);
    });
```

This script sends a request to the **local TTS server**, receives the audio response, and saves it as a `.wav` file.

---

## **Stopping the TTS Server**

* Click on the menu bar icon.
    
* Select **"Stop TTS Server"** to terminate the service.
    

---

## **Build Anything with Local TTS**

The beauty of HearItServer is its **flexibility**, it provides a **universal interface** for local TTS inference, meaning **anyone can build applications** on top of it! Some potential use cases include:

* 🤖 **AI Assistants** – Power your local AI chatbot with real-time speech synthesis.
    
* 📝 **Voice Narration** – Generate high-quality audio for videos or presentations.
    
* 🎮 **Game Development** – Implement dynamic in-game voice synthesis without cloud dependency.
    
* 🦾 **Automation** – Integrate TTS into scripts, notifications, or smart assistants.
    

With **HearItServer**, developers get **full control** over their text-to-speech processing, powered by the fastest open-source TTS model [Kokoro-82M](https://huggingface.co/hexgrad/Kokoro-82M).

---

## **Conclusion**

If you're looking for a **fast, efficient, and private** way to generate speech locally, **HearItServer** is your best bet. It harnesses the power of **Kokoro** to deliver **ultra-fast TTS inference**, making it ideal for real-world applications.

**Ready to get started? go ahead and download HearItServer and use it for your apps**

📖 **Learn more about Kokoro-ONNX:** [GitHub Repository](https://github.com/thewh1teagle/kokoro-onnx)

PS: This project is still in development and there might be bugs, expect frequent updates and improvements as I continue refining it. Feedback are always welcome!