Google’s enterprise cloud gets a music-generating AI model

On Wednesday, Google submitted updates to many artificial intelligence models generated to the media at the first end through its Cloud platform through Vertex Ai.

Lyria, the Google Text to music, is now available in a preview of the chosen customers, the company’s VEO 2 video creation model has been strengthened with the new editing and customizations options. The company has also launched a voice calming feature supported by Chirp 3, Google’s voice understanding model, for “allowing” users. Imagen 3 photo generator now offers what the company describes as “better performance”.

The updates of Cloud Next, the latest batch of Google to Corner The Enterprise Market for AI. The company may compete directly with Amazon, which offers comparable Cloud AI platform called Bedrock with a set of artificial intelligence models.

Google promotes Lyria as an alternative to kings -free music libraries. Using the model, the company said, customers can create songs in a set of styles and types, from the single-piano-jazzy mirrors to LO-Fi paths.

Meanwhile, Chirp 3 can collect speech in about 35 languages. It was first previewed earlier this year, Chirp 3 leads an immediate dedicated sound, which is supposed to reproduce a sound with 10 seconds of the sound. It is generally available. This model also supports a new inspection tool, called copies with Diazation, which separates and selects loudspeakers in records with many participants.

To prevent abuse, the immediately intended sound is subject to the “care” process to check the “appropriate use permissions”, says Google.

For VEO 2, the model can now remove background images, logos and objects from existing videos, and expand the video footage (to convert the landscape video into an image, for example). He can now adjust the corners of the camera and speed in scenes created from artificial intelligence to create timing devices, drones, and more, and can rotate between the preliminary and end -up frameworks.

These VEO features are available in the inspection at the present time.

As for the upgrades of the Imagen 3 mentioned above, Google said it is working to improve the ability of the model to remove objects and rebuild parts of lost or damaged images.

All the media created by IMAGEN, VEO and Lyria (but not the tweet) is a watermark using Google’s Synthid technology. The company said that all the models of the artificial intelligence are “compact guarantees” to protect from the creation of harmful content.

Google has not historically referred to the specific data it uses to train its models, and the technology giant stuck with that previous today. Training data tends to be a controversial topic for IP. Some companies train their models on publication and copyright businesses without obtaining permission from the rights holders. While these companies claim that the fair doctrine in the United States protects this practice, some creators do not understand understandablely. Many are fighting sellers in court.

Google Techcrunch has already been told that it provides mechanisms to cancel subscription for models training as well as compensation policy for the protection of Google Cloud and Vertex AI from prosecution’s copyright disputes.