Google’s Latest Music AI, MusicLM Can Convert Text Into Music

Google researchers have done an amazing job by developing an Artificial Intelligence(AI) that is able to create high-quality music from just text descriptions. The AI is called MusicLM which is making music creation easier than ever.

MusicLM can generate music at 24 kHz, a perfect blend of high quality and clarity that remains consistent for several minutes. This latest AI is outperforming the previous systems developed by Google both in adherence to the text descriptions as well as the audio quality.

Google published a research paper on Thursday that claims that the AI is capable of transforming a whistled or hummed melody according to the text description and also abides by the style given in the text caption. However, the tech giant is not willing to release the AI immediately as it acknowledged that there are several risks associated with the model and there is still room for improvement before it gets available for everyone.

In the research paper, Google said, “We strongly emphasize the need for more future work in tackling these risks associated to music generation — we have no plans to release models at this point.”

When it comes to the global music industry, is humongous and there are numerous styles, genres, and cultural effects in music. If this new model has to be accurate with music generation; it needs more training.

Currently, the sample generated from the AI is reflecting the biases present in the training data. According to Google, it is raising questions about the appropriateness of music generation for various cultures underrepresented in the training data as well as concerns about cultural appropriation.

The company is currently working on this AI model to make it more advanced which helps generate music more meticulously. Future works are focusing on lyrics generation, text conditioning, vocal quality, and many other aspects. In the future, MusicLM will be able to model high-level song structure along with the verse, chorus, introduction, and so on. The research paper also claimed that users can enjoy modeling the music at a higher sample rate with its improvement.

The structure of MusicLM combines three models; MuLan + AudioLM and MuLan + w2b-Bert + Soundstream. As a result, it has made more than 5,000 music-text pairs available for everyone. Clearly, Google’s new AI is embarking on a bright future for emerging music artists.

Check out For More News Updates on Google News