Spark-TTS is an advanced text-to-speech system that uses the power of large language models (LLM) for highly accurate and natural-sounding voice synthesis.
Inference Overview of Voice Cloning
Inference Overview of Controlled Generation
Generate your first AI voice clone with Spark TTS in minutes. Experience zero-shot voice cloning and multi-language support.
By using Spark TTS, you agree to our Terms of Service and Privacy Policy.
Explore the powerful features of GPT-4o mini TTS for generating human-like speech. Also available at Minitts.io.
Spark TTS is an advanced text-to-speech system powered by large language models, supporting zero-shot voice cloning and multi-language speech synthesis. It simplifies the speech synthesis process by directly reconstructing audio from LLM-predicted codes and delivers natural and fluent speech results. Spark TTS supports both Chinese and English, offers controllable speech generation through adjustable parameters like gender, pitch, and speaking rate, and is suitable for both research and production environments.
Spark TTS is built entirely on Qwen2.5, eliminating the need for additional generation models like flow matching. It directly reconstructs audio from the code predicted by the LLM, streamlining the process and improving efficiency.
Spark TTS supports zero-shot voice cloning, which means it can replicate a speaker's voice even without specific training data for that voice. This is ideal for cross-lingual and code-switching scenarios.
Spark TTS supports both Chinese and English, and is capable of zero-shot voice cloning for cross-lingual and code-switching scenarios, enabling the model to synthesize speech in multiple languages with high naturalness and accuracy.
Spark TTS supports creating virtual speakers by adjusting parameters such as gender, pitch, and speaking rate.
Spark TTS enables high-fidelity speech reconstruction at low bitrates, delivering natural and fluent speech synthesis results.
Spark TTS offers both Web UI and CLI interfaces, supports multiple operating systems, simple installation and deployment, and easy integration into various application scenarios.
Discover our codebase, models, and documentation to get started with Spark TTS.
Have a different question and can't find the answer you're looking for? Reach out to our support team by sending us an email and we'll get back to you as soon as we can.
Spark TTS is an advanced text-to-speech system powered by large language models (LLMs). It leverages the Qwen2.5 model to directly reconstruct audio from predicted codes, eliminating the need for additional acoustic feature generation models. This streamlined approach enhances efficiency and delivers high-quality, natural-sounding speech synthesis.
Spark TTS offers several key features: high efficiency through direct audio reconstruction, zero-shot voice cloning for replicating voices without specific training data, bilingual support for Chinese and English, controllable speech generation via adjustable parameters (gender, pitch, speaking rate), high-quality voice synthesis with natural and fluent results, and flexibility with both Web UI and CLI interfaces.
You can install Spark TTS by cloning the repository from GitHub and following the provided installation guide. Once installed, you can use it via command line or Web UI to perform voice cloning and speech synthesis tasks.
Yes, Spark TTS allows you to create custom voices by adjusting parameters such as gender, pitch, and speaking rate. This controllability enables the generation of diverse and tailored virtual speakers for various applications.
Spark TTS delivers high-fidelity voice synthesis even at low bitrates. It produces natural and fluent speech that closely mimics human speech patterns, making it suitable for applications where high-quality audio is essential.
Yes, Spark TTS supports both Chinese and English. It can seamlessly switch between languages and maintain natural speech synthesis, making it a versatile tool for multilingual applications.
Spark TTS is designed to be efficient, flexible, and powerful for both research and production environments. Its high-quality voice synthesis and controllable features make it a valuable tool for commercial applications such as voice assistants, content creation, and customer service solutions.
Spark TTS can be run on systems with Python 3.8 or higher. For faster inference speeds, a CUDA-supported GPU is recommended. The exact requirements may vary depending on the specific use case and scale of operations.
To optimize Spark TTS performance, ensure you have the correct hardware setup, such as a compatible GPU. Additionally, follow the installation guide carefully and utilize the provided configuration options to tailor the system to your specific needs.
You can visit the official Spark TTS website at https://sparkaudio.github.io/spark-tts/ for detailed documentation, demos, and resources. The GitHub repository at https://github.com/SparkAudio/Spark-TTS also contains the source code and additional information.