Orpheus-TTS-Local

A standalone Text-to-Speech application using the Orpheus TTS model with a modern Gradio interface.

UI Screenshots:

Features

• 🎧 High-quality Text-to-Speech using the Orpheus TTS model
• 💻 Completely standalone - no external services or API keys needed
• 🔊 Multiple voice options (tara, leah, jess, leo, dan, mia, zac, zoe)
• 💾 Save audio to WAV files
• 🎨 Modern Gradio web interface
• 🔧 Adjustable generation parameters (temperature, top_p, repetition penalty)
• 😊 Emotive speech generation with natural expressions

Sample Audio

Listen to a sample of the generated speech: Sample Audio

Quick Setup

1. Install Python 3.8 or higher
2. Install dependencies:

   python3 -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate
   pip install -r requirements.txt

3. Run the application:

   python gradio_orpheus.py

The application will automatically:

• Download the Orpheus TTS model on first run
• Download and initialize the SNAC audio codec
• Start the Gradio web interface

Usage

1. Open your web browser and navigate to the URL shown in the terminal (usually http://127.0.0.1:7860)
2. Enter the text you want to convert to speech
3. Select a voice from the dropdown menu
4. Adjust generation parameters if desired:
   • Temperature: Controls randomness (0.0-1.0)
   • Top P: Controls diversity (0.0-1.0)
   • Repetition Penalty: Controls repetition (1.0-2.0)
5. Click "Generate Speech" to create the audio
6. Play the generated audio directly in the browser or download it

Sample Prompts

Regular Text-to-Speech

Welcome to our presentation. Today, we'll be discussing the latest developments in artificial intelligence and machine learning.

Emotive Text-to-Speech

<giggle>Oh, that's hilarious!</giggle> I can't believe what just happened. <laugh>This is the funniest thing I've seen all day!</laugh>
<sigh>But seriously though,</sigh> we need to focus on the task at hand. <gasp>Look at what we've accomplished!</gasp>

Available Voices

• tara - Best overall voice for general use (default)
• leah
• jess
• leo
• dan
• mia
• zac
• zoe

Emotion Tags

You can add emotion to the speech by adding the following tags:

<giggle>
<laugh>
<chuckle>
<sigh>
<cough>
<sniffle>
<groan>
<yawn>
<gasp>

Technical Details

This implementation:

• Uses llama-cpp-python to run the Orpheus model locally
• Uses the SNAC neural audio codec for high-quality audio generation
• Processes tokens in chunks of 28 for optimal audio quality
• Supports both CPU and GPU (CUDA/MPS) acceleration

Requirements

• Python 3.8 or higher
• 8GB RAM minimum (16GB recommended)
• CUDA-capable GPU (optional, for faster generation)
• See requirements.txt for Python package dependencies

License

Apache 2.0