Tool not found.
Back to Home
Open-source toolkit for speech, audio, and music AI research featuring text-to-speech, voice cloning, speech recognition, singing synthesis, and audio generation.
Amphion is an open-source audio, speech, and music foundation model toolkit developed by the OpenMMLab community. It provides a unified platform for building, training, fine-tuning, and deploying state-of-the-art AI models across speech synthesis, speech recognition, voice conversion, singing voice synthesis, music generation, audio generation, and multimodal audio applications. Amphion integrates modern foundation models, supports distributed training, offers reproducible research pipelines, and includes pretrained checkpoints for numerous speech and music tasks. Researchers and developers can experiment with large-scale audio models, customize training workflows, and build production-ready speech AI systems using PyTorch-based infrastructure.
