Google has upgraded Spaces, the creative tool inside its Flow Music platform, adding the ability to generate full songs, edit them, split them into stems, and create images, lyrics, and video alongside the audio. The update was announced via the Flow Music account on X on July 24, 2025 and detailed by Music Business Worldwide on July 29, 2026.

Spaces was originally pitched as a "vibe coding" tool — a way for artists to build custom instruments and effects using plain language rather than code. When Google acquired the platform in February 2025, it was still called ProducerAI. Google rebranded it to Flow Music in April 2025 and folded it into Google Labs under the oversight of Elias Roman, Senior Director of Product Management. The platform traces its roots back to Riffusion, an open-source project that went viral in 2022.

The new song features are substantial. Users can now create covers of generated tracks, replace specific sections of a song, and extend tracks beyond their original length. Stem splitting — separating a finished recording into individual channels like vocals, drums, and bass — gives users the building blocks to remix or rearrange. Image, lyric, and video creation capabilities round out the update, turning Spaces from a sound-design tool into something closer to a full production environment.

Under the hood, Flow Music runs on Lyria 3 Pro, a music-generation model built by Google DeepMind. The platform also draws on Google's Gemini model for chat, Veo for video generation, and a model called Nano Banana for image generation. At Google I/O in May 2026, the company announced that Flow Music users can use Gemini Omni — Google's conversational agent — to direct the creation of shareable music videos through dialogue, as confirmed in Google's I/O 2026 announcement.

Every piece of content produced through Flow Music carries a SynthID watermark, Google's system for identifying AI-generated material. That matters in a music industry still working out how to distinguish human-made recordings from machine-generated ones, particularly as streaming platforms and rights holders push for transparency around AI content.

Google is not building Flow Music in isolation. In May 2025, the independent music company Believe partnered with Google to offer Flow Music to artists across its distribution network and its self-release arm TuneCore, giving the tool a direct pipeline to working musicians who use those services to release music. The Spaces additions also bring Flow Music into closer competition with Suno and Udio, two AI music platforms that have already built their own editing and production tools.

The official Google Labs homepage describes Flow Music as "your music studio for creating original songs and music videos with an AI producer," and the company hosts a help section covering song creation, Spaces management, and getting started at labs.google.

For now, the tool exists in a space where the technology is moving faster than the policy around it. Google has placed its watermarking bet. Whether the industry treats a SynthID tag as sufficient disclosure — and whether artists offered Flow Music through TuneCore and Believe actually use it — are separate questions that no single update can answer.