Google has introduced a new AI tool ‘MusicLM’ that can generate high-quality music from text descriptions. MusicLM uses a hierarchical sequence-to-sequence modeling approach to generate 24 kHz music from text description and the music at 24 kHz. Google also claims that MusicLM outperforms previous systems.
This tool takes text or an existing melody as input. It means this tool can also convert whistled and hummed melodies into audio tracks. On the dedicated Github page, Google has also added several demo auto tracks to give you an idea of how this AI tool works.
MusicLM has been trained on a dataset of 280,000 hours of music. I heard several audio tracks created by this online tool and it feels that these tracks were composed by human artists. With time, this AI tool will become even better.
Google has also shown how MusicLM can create music tracks by taking a combination of pictures and captions as input. One can also set the experience level of the AI.
One of the major concerns of this tool is its tendency to incorporate copyrighted material from training data into the generated songs. If this tool becomes available for all internet users, it will surely suffer legal issues.
