MusicLM

MusicLM is an AI tool that utilizes given prompts and datasets to produce high-quality music text. It employs a hierarchical sequence-to-sequence model to generate consistent music for extended durations, surpassing previous systems in terms of sound excellence. It can also modify melodies based on textual prompts and generate audio by employing various tests and samples. Furthermore, MusicLM offers a dataset containing 5.5k music-text pairs enriched with detailed textual explanations from human specialists.