Service status: Google has closed the Music AI Sandbox trusted-tester program and now directs applicants to ProducerAI (joined Google Labs in February 2026 and renamed Google Flow Music in April 2026). No official announcement gives an exact shutdown date. This article is kept as a record of the tool as it was. Unrelated sites use the ProducerAI name, so start from Google’s own announcement.
An experimental music generation tool that Google DeepMind is developing in collaboration with musicians. Beyond generating music samples from text descriptions, it can extend existing sounds forward or backward and transform the mood or genre of specific sections — the emphasis is not just on “creating from scratch” but on “developing works in progress.” Born from collaborations with artists in YouTube’s Music AI Incubator, it was introduced to the public in 2024, and in April 2025 it was upgraded with the Lyria 2 music generation model, expanded features, and broader access for musicians, producers, and songwriters in the U.S. All generated music carries a SynthID watermark.
Key Features
- Create — generate music samples from text: Describe genres, moods, vocal styles, and instrumentation in words to generate multiple music samples. You can also place your own lyrics on a timeline and specify tempo and key
- Extend — generate continuations of a sound: The AI generates continuations of uploaded audio or generated clips — useful for exploring how a track could develop, or for breaking through creative blocks
- Edit — transform mood and genre: Transform the mood, genre, or style of an entire clip or a specific section with text instructions. You can also try preset transformations and blend clips to create transitions
- Powered by Lyria 2 and Lyria RealTime: In addition to the high-quality Lyria 2 music generation model, it offers Lyria RealTime for interactive, real-time music creation
- SynthID watermarking: Every piece of generated music carries an invisible watermark, making AI-generated content identifiable
- Co-developed with artists: Built with feedback from professional musicians through YouTube’s Music AI Incubator, with a design that reflects real production workflows
Pricing
| Plan | Price | Key Points |
|---|---|---|
| Experimental access (waitlist) | Not disclosed | Rolling out to musicians, producers, and songwriters in the U.S. |
Pricing information is current as of August 2026. No pricing plans have been published, and access requires joining a waitlist. Check the official site for the latest information.
Pros & Cons
✅ Pros
- Covers not only generation but also extension and editing, making it strong for developing works in progress
- Designed as a production aid rather than a one-shot song generator, with lyric placement, tempo, and key controls
- Early access to Google DeepMind’s latest music models (the Lyria family)
- SynthID watermarking ensures transparency about AI-generated content
⚠️ Cons
- Waitlist-based and U.S.-focused, so it may not be immediately available from other countries such as Japan
- As an experimental tool, features and availability may change without notice
- Commercial-use terms and pricing are not specified, making it hard to plan business use
- Compared with services that generate finished songs in one shot, some music production knowledge helps to get the most out of it
Comparison with Similar Services
| Comparison | Music AI Sandbox | Suno | Udio | Stable Audio |
|---|---|---|---|---|
| Provider | Google (DeepMind) | Suno | Udio | Stability AI |
| Main use | Generating, extending, and editing production material | Full songs with vocals | Full songs with vocals | BGM and sound generation |
| Availability | Experimental (waitlist) | Generally available | Generally available | Generally available |
| Pricing | Not disclosed | Free tier / paid plans | Free tier / paid plans | Free tier / paid plans |
| Strength | Extending and partially editing works in progress | Ease of use and polish | Sound quality and genre range | Fine control for audio production |
Who Is It For
- Musicians and producers who want to bring AI into songwriting and arrangement ideation
- People who want AI as a tool to develop their own material, rather than an automatic song generator
- Anyone eager to try Google DeepMind’s latest music generation models early
- Creators who value transparency (watermarking) in AI-generated music
Summary
Music AI Sandbox is an experimental, production-oriented music AI tool that pairs text-based generation with extension and editing. Its direction differs from services that mass-produce finished songs — it is designed as an instrument for musicians to develop their own ideas. The waitlist system and limited regional availability are hurdles, but its features keep expanding alongside the evolution of the Lyria model family, making it a service worth watching for anyone who wants to bring AI into music production.