EmotiVoice is an open-source, multi-voice, and prompt-controlled Text-to-Speech (TTS) engine supporting English and Chinese. It excels at emotional synthesis, offering over 2000 distinct voices and a wide range of controllable emotions, making it suitable for researchers and developers building advanced voice applications.

How It Works

EmotiVoice leverages a prompt-based control mechanism, allowing users to influence speech characteristics like emotion and style through text prompts. This approach, inspired by PromptTTS, enables fine-grained control over the synthesized output beyond traditional speaker identity, offering a more expressive and nuanced TTS experience.

Quick Start & Requirements

• Docker: docker run -dp 127.0.0.1:8501:8501 syq163/emoti-voice:latest (requires NVIDIA GPU and NVIDIA Container Toolkit).
• Full Install: Requires Python 3.8, PyTorch, torchaudio, and several other Python libraries. Pretrained model files must be downloaded separately.
• Demo: A web UI is available at http://localhost:8501 after running the Docker image or streamlit run demo_page.py after installation.
• API: An OpenAI-compatible TTS API is available at http://localhost:8000/.
• Docs: Wiki Page

Highlighted Details

• Supports over 2000 voices and controllable emotional synthesis (happy, sad, angry, etc.).
• Includes voice cloning capabilities with provided recipes.
• Offers an OpenAI-compatible TTS API for easier integration.
• Provides both a web UI and a scripting interface for batch generation.

Maintenance & Community

The project is actively developed, with recent updates including speed tuning, a Mac app, and an HTTP API. Community engagement is encouraged via GitHub issues and a WeChat group.

Licensing & Compatibility

EmotiVoice is licensed under the Apache-2.0 License, permitting commercial use and integration with closed-source projects. The interactive page has a separate User Agreement.

Limitations & Caveats

Currently, EmotiVoice primarily focuses on pitch, speed, and energy for style control, not explicitly gender. Support for additional languages like Japanese and Korean is under development.