ZyphraAI recently released its latest multilingual text-to-speech (TTS) model, **Zonos-TTS**, which is fully open-source and commercially usable under the **Apache 2.0 License**. Zonos-TTS not only supports real-time voice cloning but has also been trained on 200,000 hours of English voice data, demonstrating outstanding performance.
Zonos-TTS offers two deployment options: local deployment and convenient API services. For regular users, ZyphraAI provides a service to generate **100 minutes** of audio for free each month; professional users can choose a package of **300 minutes for $5**, with additional usage charged at **$0.02 per minute**. Notably, the real-time voice cloning feature is completely free and supports high audio quality and emotional control, providing users with great flexibility and practicality.
Additionally, Zonos-TTS supports audio prefix input, further enhancing its application scenarios. According to ZyphraAI's tests, on an **RTX 4090** graphics card, the model can run at approximately **twice the real-time speed**, showcasing its high efficiency. This open-source release provides developers and enterprises with a powerful and flexible tool to advance multilingual real-time voice cloning technology.