Google Labs adds ProducerAI to build full songs with DeepMind models

Most AI music tools today treat song creation like a digital slot machine: you type in a genre, pull the lever, and hope the machine spits out a hook that does not sound muddy.

On Tuesday, Google Labs brought generative music platform ProducerAI into its fold. Backed by The Chainsmokers and developed with artists like Grammy-winning rapper Lecrae, the tool is aimed at bridging the gap between casual prompt engineering and actual track production.

Under the hood, ProducerAI taps a preview version of Google DeepMind's Lyria 3 music generation model, alongside Gemini, Nano Banana, and the Veo video model. Every track exported from the platform carries Google's invisible SynthID watermark.

Granular knobs over slot-machine prompts

Last week, Google added Lyria 3 into the Gemini app for general consumers. ProducerAI, however, is aimed at musicians who actually care about the mix.

Users can still start with simple text prompts like "make a lofi beat," but the system allows granular adjustments over tempo, low-end punch, reverb throws, and time-aligned lyric placement. DeepMind's Jeff Chang noted that it is built for careful curation rather than pressing a button a hundred times.

One standout addition is Spaces. The feature lets creators use natural language to generate custom virtual instruments and audio effects, ranging from basic keyboards to node-based modular patch bays that other users can remix.

The platform is available globally on its website with both free and paid tiers.

Why it matters

For bedroom producers and content creators, this shifts generative music from novelty demos to a flexible sandbox. If you need custom stems, want to audition an instrument in an existing mix like Wyclef Jean did with Lyria, or need bespoke background tracks, it is definitely worth trying today.

For the music tech landscape, Google is taking direct aim at tools like Suno. While Suno has already produced viral Billboard-charting hits and landed users multimillion-dollar record deals, Google is betting on deep editorial control rather than just full-track automation.

We think positioning this as a human collaborator is a savvy defensive play. The music industry is actively fighting generative audio, from the 2024 open letter signed by hundreds of major stars to the recent $3 billion copyright lawsuit against Anthropic. By getting credible artists onboard and emphasizing creative control, Google is trying to dodge the backlash.

The lingering question, however, is training data. While courts debate the legality of scraping copyrighted music, no amount of SynthID watermarking will fully ease the industry's anxiety if AI-assisted tracks start eating into streaming royalties on Spotify.