SpeakItAI – Neural Text-to-Speech with Azure & Gradio

Convert text to speech using Microsoft Azure Neural Text-to-Speech (TTS) and a simple Gradio web interface.

---

A simple, interactive interface for converting your text to realistic speech.

🎯 Features

• Neural TTS with dynamic voice selection across 140+ Azure-supported languages
• Adjustable speaking style, rate, and pitch
• Input via textbox or upload a .txt file
• Output as .wav file, played directly in the browser
• Dropdowns now auto-populate with default language, voice, and style
• Human-readable language names in the UI (e.g., "English (UK)" instead of en-GB) — falls back to code if name not defined
• 🔧 Manage display names and visible languages directly in the UI – no need to edit files or restart the app
• Modular architecture – ready for expansion

---

🆓 Azure Free Tier

Microsoft Azure offers 500,000 characters per month free for Neural Text-to-Speech on the F0 (free) pricing tier.

• ⚡ Billing is per character
• ✅ Free quota resets monthly
• 🧪 No credit card required to start

📖 More info: Azure Speech Pricing

---

🌍 Language and Voice Support

Azure Neural TTS supports 140+ languages and dialects, with many realistic male and female voices, including:

• 🇬🇧 British English
• 🇺🇸 American English
• 🇫🇷 French
• 🇩🇪 German
• 🇷🇺 Russian
• 🇨🇳 Chinese
• 🇪🇸 Spanish
• 🇮🇳 Hindi
• 🌐 And more

📖 Full voice list and styles:
👉 Azure Language & Voice Support

---

🚀 Setup Instructions

1. Clone the Repository

git clone https://github.com/loglux/SpeakItAI.git
cd SpeakItAI

🔐 Azure Setup (Required)

Before running this app, you need an active Azure Speech resource.

1. Go to the Azure Portal
2. Create a Speech resource (Free F0 tier available)
3. Copy the Key and Region from the resource's "Keys and Endpoint" section
4. You will paste them into a .env file as shown below:

2. Create .env File

cp .env.example .env

Then fill in your Azure credentials:

AZURE_KEY=your_azure_key
AZURE_REGION=your_azure_region

💡 Example region: ukwest, eastus, westeurope, etc.

---

3. Install Dependencies

Using virtual environment (recommended):

python -m venv venv
source venv/bin/activate   # On Windows: venv\Scripts\activate
pip install -r requirements.txt

---

4. (Optional) Update Voice List from Azure

To fetch the latest voices and update config.json, run:

python tts/azure/update_config.py

---

5. Run the App

python app.py

---

📝 Usage Notes

• If both textbox and file are provided, the file takes priority.
• Only .txt files are accepted for upload.
• The output is saved and played as a .wav file.
• If a voice does not support styles, "default" will be used automatically.

---

📂 Project Structure

SpeakItAI/
├── app.py
├── .env.example
├── requirements.txt
├── README.md
├── audio_outputs/                # automatically created on runtime
│
├── screenshots/
│   └── interface.png             # UI preview
│
└── tts/
    ├── __init__.py
    ├── base.py                   # optional: abstract provider interface
    └── azure/
        ├── __init__.py
        ├── core.py               # AzureTTS implementation
        ├── config.py             # language label loading and utilities
        ├── config.json           # auto-generated voice config from Azure
        ├── language_labels.json  # editable language name mappings (UI dropdowns)
        └── update_config.py      # fetch and build config.json

---

🧩 Configuration Notes

Voice Configuration (config.json)

• Run tts/azure/update_config.py to fetch the latest Azure voice data.
• This generates a new tts/azure/config.json with all supported languages, voices, genders, and available styles.
• The app reads from config.json at runtime to populate the voice and style options.

Language Filter (config.py)

Language Labels (language_labels.json)

• The list of display names for languages** is stored in a separate file:
  tts/azure/language_labels.json.
• You can add, rename, or delete languages using the “Edit Languages” tab in the UI — no need to edit files manually or restart the app.
• This list controls which languages appear in the dropdown, and how they are named (e.g., "English (UK)" instead of "en-GB").

Example structure:

{
  "en-GB": "English (UK)",
  "ru-RU": "Russian",
  "ar-KW": "Arabic (Kuwait)"
}

Behaviour if the file is empty or missing:

• The app will display all available languages found in config.json.
• These will be shown using their locale codes only (e.g., "fr-FR", "hi-IN", "de-DE"), without readable labels.

💡 Use the “Edit Languages” tab to selectively re-add friendly names for only the languages you want to appear.

---

🨠 Architecture Note

The codebase is modular and ready for extension:

• Add new languages or accents in tts/config.py
• Replace the interface with FastAPI or Flask without touching core.py
• Support alternative providers like ElevenLabs, Bark, or Google Cloud TTS later

---

🛡 License

This project is licensed under the MIT License.