Vertex AI Google Media Editing
- JAKARTA – Google has augmented its Vertex AI platform with several new service models, including Lyria, an AI music generator.
- The new Vertex AI capabilities enable users to construct complete production assets.
- Lyria, currently available in a limited preview, is designed to produce high-fidelity audio and capture musical nuances across various genres.
Google Enhances Vertex AI with Music Generation and Expanded Media Capabilities
Table of Contents
- Google Enhances Vertex AI with Music Generation and Expanded Media Capabilities
- Google Vertex AI: Your Questions Answered
- What is Google vertex AI?
- What are the new features added to Google Vertex AI?
- What is Lyria, and how does it work?
- What are the benefits of using Lyria?
- How has the Veo 2 video model been updated?
- What’s new with Audio Chirp 3?
- What improvements were made to Image model 3?
- Is my data secure when using Vertex AI?
- Summary of Vertex AI Features
JAKARTA – Google has augmented its Vertex AI platform with several new service models, including Lyria, an AI music generator. According to Warren Barkley, Google’s Senior Director of Product management, this addition makes Vertex AI the first platform to offer generative models for video, images, sound, and music. Barkley announced the update in a statement released Wednesday, April 9, 2025.
Vertex AI Expands Production Asset Creation
The new Vertex AI capabilities enable users to construct complete production assets. The service accepts text prompts to generate images and video assets, complete with music and sound.
Lyria: AI Music Generation
Lyria, currently available in a limited preview, is designed to produce high-fidelity audio and capture musical nuances across various genres. Google says Lyria can accelerate video content creation, podcast production, and brand campaigns without requiring additional music licenses.
Users can provide detailed prompts, such as: “Create high-energy bebop music, prioritizing saxophone and trumpet solos with complex, high-speed phrases.”
The prompt can be further refined with instructions like: “Add a rhythmic piano accompaniment with walking bass and fast drums. The atmosphere should be thrilling and intense, capturing the nuances of a smoky jazz club, accentuating virtuosity and improvisation. The listener should be unable to sit still.”
Veo 2 Updates: Advanced Video Editing
Google also updated other generative models in Vertex AI. Veo 2, the advanced video model, now supports editing features like inpainting for automatic background element removal and outpainting to expand video frames for different screen formats.
Veo 2 also includes automatic camera control features, such as presets for camera movements, time-lapse effects, and drone-style shots. An interpolation feature allows users to connect two video clips with smooth transitions. These features are available in a preview version.
Audio Chirp 3: Custom Voice and enhanced transcription
The Audio chirp 3 model has also been enhanced. the Instant Custom Voice feature allows users to create realistic custom voices from just a 10-second voice recording. this feature is intended for personal services such as call centers and voice branding.
Chirp 3 also includes Transcription with Diarization,which distinguishes and identifies speakers in a single voice recording. Chirp 3 supports more than 35 languages and provides eight high-quality (HD) sound options.
Image Model 3: improved Image Repair
Google has improved the quality of Image Model 3, equipping it with more accurate inpainting capabilities to repair damaged or missing parts of images. Object removal is also more natural and seamless than in previous versions.
Commitment to Security and Ethics
Google states that all developments are designed with strict security and ethical principles. Deepmind’s SynthID digital watermarking technology is automatically applied to each frame of images, videos, and audio produced by Images, Veo, and Lyria to prevent disinformation and abuse.
All models are equipped with security filters to prevent the creation of dangerous content. A data governance system ensures customer data is not used to train AI models. According to Barkley, users can safely use the content produced because Google will protect them from third-party IP claims, including copyright.
Google Vertex AI: Your Questions Answered
What is Google vertex AI?
Google Vertex AI is a unified machine learning (ML) platform designed to help businesses build, deploy, and scale ML models quickly. It offers a comprehensive suite of tools and services for the entire ML lifecycle, from data preparation and model training to deployment and monitoring.The platform has recently been enhanced with new service models, including capabilities for generating music, video, images, and more.
What are the new features added to Google Vertex AI?
Google has expanded Vertex AI’s capabilities with several new service models, as announced on April 9, 2025. These include:
Lyria: An AI music generator.
Veo 2: An advanced video model.
Audio Chirp 3: Enhanced custom voice and transcription features.
Image Model 3: Improved image repair capabilities.
Warren Barkley, google’s Senior Director of Product management, announced the updates.
What is Lyria, and how does it work?
what is Lyria?
lyria is an AI music generator integrated into Google’s Vertex AI platform. It’s designed to create high-fidelity audio that can capture musical nuances across various genres. Lyria is currently available in a limited preview version.
How can I use Lyria?
Users can prompt Lyria with detailed text-based descriptions to generate music. For exmaple, you could specify: “Create high-energy bebop music, prioritizing saxophone and trumpet solos with complex, high-speed phrases.”
You can further refine your prompt, as an example, with instructions such as: “Add a rhythmic piano accompaniment with walking bass and fast drums. The atmosphere should be thrilling and intense, capturing the nuances of a smoky jazz club, accentuating virtuosity and improvisation.The listener should be unable to sit still.”
What are the benefits of using Lyria?
Lyria offers several advantages:
Accelerated Content Creation: It can speed up video content creation, podcast production, and brand campaigns.
Reduced Licensing concerns: It enables you to create music without the need for additional music licenses.
Versatile music Generation: It can produce music across various genres, allowing for creative adaptability.
How has the Veo 2 video model been updated?
The Veo 2 video model, within Vertex AI, has received some key updates, including:
Inpainting and Outpainting: These features allow for automatic background element removal and expanding video frames for different screen formats.
Camera Control: Includes presets for camera movements, time-lapse effects, and drone-style shots.
Interpolation: Connects two video clips with smooth transitions.
These features are all available in a preview version.
What’s new with Audio Chirp 3?
The Audio Chirp 3 model has been enhanced with:
Instant Custom Voice: Creates realistic custom voices from only a 10-second voice recording, useful for personal services like call centers and voice branding.
Transcription with Diarization: Distinguishes and identifies speakers within a single voice recording.
Language Support: supports over 35 languages.
High-Quality Sound Options: Provides eight high-quality (HD) sound options.
What improvements were made to Image model 3?
Image Model 3 has been upgraded to:
Improved Inpainting: Offers more accurate capabilities to repair damaged or missing parts of images.
Seamless Object Removal: Makes object removal more natural than previous versions.
Is my data secure when using Vertex AI?
Yes. Google is committed to security and ethics with all Vertex AI developments.
What security measures are in place?
Google implements several security measures:
SynthID Watermarking: DeepMind’s SynthID digital watermarking technology is automatically applied to images, videos, and audio produced by Images, Veo, and Lyria to prevent disinformation and abuse.
Security Filters: All models are equipped with security filters to prevent the creation of hazardous content.
* Data Governance: A data governance system ensures customer data is not used to train AI models.
Can I use the content produced by Vertex AI safely?
According to Warren Barkley, users can safely use the content because google will protect them from third-party IP claims, including copyright.
Summary of Vertex AI Features
Here’s a summary of the new features in a structured HTML table:
| Feature | Description | Primary Use Cases |
|---|---|---|
| lyria | AI Music Generation | Video content creation, podcast production, brand campaigns |
| Veo 2 | Advanced Video Editing | Inpainting, outpainting, camera control, transitions |
| Audio Chirp 3 | Custom Voice and Enhanced Transcription | Call centers, voice branding, transcribing audio with speaker identification |
| Image Model 3 | Improved Image Repair | Repairing damaged images, seamless object removal |