Video, image, speech and music. Google Cloud () claimed the first company title to provide AI models for creating all types of media on a platform. This Wednesday (9), the company included Lyria, a model that turns text commands into music, to its vertex catalog AI.
Also read:
The new model promises to produce “high-default audios, meticulously capturing nuances and delivering rich and detailed compositions by a wide series of genres.”
In a demonstration made at the Google Cloud Next 25 ′ event, a song with subjective descriptions such as “capturing the late night feeling in a smoky jazz club” as well as objective guidelines such as “prioritizing dizzying saxophone and trumpet soils”.
Also read:
“Today we make Lyria available on Google Cloud to transform prompts of text in clips 30 seconds music tracks and we are the first hyperscaler [grandes empresas de processamento de dados em nuvem] Offering this possibility, ”said Google Cloud CEO Thomas Kurian in a presentation at the event.
Continues after advertising
According to Google Cloud, the tool should quickly lead companies to create soundtracks to marketing campaigns, product launches, or immersive experiences. You must also meet video production, podcasts and digital content creation.
Video, Image and Voice
In a test display of its AI models for media creation, Google showed the classic “The Wizard of Oz” (1939) on “The Sphere”, a giant spherical screen in Las Vegas. Scenes from the original movie were reconstituted and pieces that escaped the original dimension in which it was filmed were created from scratch. A shot from the original showing the head of protagonist Dorothy (played by Judy Garland), for example, was included a dancing body to fill the screen dimensions.
In addition to Lyria, VEO 2, video creation, Chirp 3, audio, and Imagen 3 models are also available on AI VERTEX, Google Cloud Language Model Training Platform. Tool updates were also announced at the event.
Continues after advertising
Also read:
Veo 2 now has a camera control option. Users can, an example, climb an image in the app and ask him to anime with a certain movement, such as a drone flyfish.
The Chirp 2 update allows you to create custom voices based on 10 seconds of audios sent to the model. It is also possible to add narrations to existing recordings and transcribe talk with automatic identification of the speakers.
Continues after advertising
By Imagen 3, users can now touch up and reconstruct missing or damaged areas of an image.
Among the partners, Adobe closed a partnership for the inclusion of Imagen 3 models on the Adobe Express platform; Kraft Heinz, L’Oreal and Bending Spoons already use Vertex AI tools in campaign creation, clips and photos.
*The reporter traveled to Las Vegas at the invitation of Google Cloud
