Amphion is an open-source toolkit by OpenMMLab for audio, music, and speech generation research and production. It covers a broad range of generation tasks including text-to-speech, voice conversion, singing voice synthesis, text-to-audio, and vocoder training, with support for architectures like VALL-E, NaturalSpeech2, MaskGCT, and Vevo for zero-shot capabilities. Pre-trained models are available on HuggingFace and ModelScope, making it accessible for both researchers and engineers.