Local11Labs: High-Quality Text-to-Speech & Podcast Generator

Local11Labs is a powerful text-to-speech and podcast generation tool powered by the lightweight Kokoro-82M model. Generate natural-sounding speech and multi-speaker podcasts locally on your machine.

Key Changes

1. Gradio WebUI Integration

   • Integrated Gradio WebUI into the application to enhance user interaction and accessibility.

2. Text-to-Speech (TTS) Tab

   • Input Options: Users can provide input via a text field or upload a .txt file.
   • Speech Rhythm Control: Adjust the rhythm and pacing of generated speech.
   • Voices Dropdown Menu: Select from a variety of available voice profiles.
   • Device Selection: Automatic device detection with the option to specify CUDA or CPU.

3. Podcast Dialogue Generation Tab (Powered by Gemini)

   • Enables dynamic generation of podcast dialogues.
   • Input customizable host names to create unique and personalized conversations.

4. Podcast Audio Generation

   • Script Input Options:
     • Use the JSON output from the Dialogue Generation tab.
     • Upload a custom JSON file.
     • Directly edit or input text within the interface.
   • Dynamic Host Mapping:
     • Assign available voices to host names.
     • Add as many hosts as needed without any limitations.
   • Speech Rhythm Control: Fine-tune the speech rhythm for each host to enhance the audio experience.

Key Features

• ⚡ Fast and efficient text-to-speech generation
• 🎙️ Multi-speaker podcast creation with distinct voices
• 📝 Smart text chunking for handling long content
• 🎛️ Customizable voice profiles with caching
• 🔊 Professional audio quality with natural pauses and transitions
• 🚀 Easy to use with minimal setup required

Podcast demo

https://github.com/user-attachments/assets/fdb09bfb-48d6-4083-852d-c4353b4df2a9

Quick Start

https://github.com/user-attachments/assets/3bc23c3f-86f0-4aad-a34a-6a16cc3aa80a

Text-to-Speech Demo

Try out basic text-to-speech generation in our interactive Colab notebook:

Podcast Generation Demo

Create multi-speaker podcasts using our podcast generation notebook:

Webui demo :

Notebook will be added soon for not you can clone it and run it.

TODO List

• [ ] Add REST API endpoints for remote TTS generation
• [ ] Implement streaming audio support
• [x] Create web interface for easy usage
• [ ] Add Docker support for easy deployment

Screenshots :