MeloTTS is a high-quality multilingual text-to-speech library developed collaboratively by MIT and MyShell.ai. It delivers fast, natural-sounding audio synthesis across multiple languages and regional accents including American, British, Indian, and Australian English, as well as Spanish, French, Chinese (with mixed Chinese-English support), Japanese, and Korean. A standout feature is its performance: MeloTTS is fast enough for CPU real-time inference, enabling practical deployment without specialized GPU hardware. Built upon established TTS architectures including VITS, VITS2, and Bert-VITS2, it delivers high-fidelity audio output across all supported languages. The library is released under the MIT license, making it free for both commercial and non-commercial use. MeloTTS provides three usage pathways: quick usage without installation, local installation and setup, and custom dataset training for specialized applications. With 7,200+ stars on GitHub and backing from both academic research at MIT and MyShell.ai's production expertise, MeloTTS represents a practical, accessible solution for developers needing multilingual speech synthesis capabilities.