LiveKit Orpheus TTS Plugin

A high-performance text-to-speech plugin for LiveKit agents using the Orpheus TTS model, featuring both streaming and non-streaming implementations.

Features

• Low Latency: ~160ms time-to-first-byte (TTFB) on RTX 4090
• Streaming Support: Real-time audio generation with streaming capabilities
• Multiple Voice Options: Support for various fine-tuned voice models
• LiveKit Integration: Seamless integration with LiveKit agents framework

Requirements

• LiveKit Agents v1.2 or higher
• Orpheus streaming server (recommended for best performance)

Installation

1. Clone or download this plugin into your LiveKit-based agents project root directory
2. Ensure you have the Orpheus streaming server running

Usage

Initialize your agent session with the LocalTTS plugin:

from your_plugin_path import LocalTTS

session = AgentSession(
    # ... other configuration
    tts=LocalTTS(
        base_url="<orpheus_server_url>",  # e.g., "http://localhost:8000"
        voice="tara",  # or any other available voice
    )
)

Fine-tuned Models

I have finetuned Orpheus on various custom voices, all are availabe on my huggingface:

• Maya Voice: taresh18/orpheus-maya
• Lucy Voice: taresh18/orpheus-lucy
• Anime Speech: taresh18/orpheus-animespeech
• AnimeVox Dataset: taresh18/AnimeVox

Performance

• Latency: ~160ms TTFB on RTX 4090 GPU
• Streaming: Real-time audio generation and playback
• Compatibility: Tested with LiveKit Agents v1.2