Google Lyria 2

Released in April 2025, Google Lyria 2 is an advanced text-to-music model from Google DeepMind that generates high-fidelity instrumental music. It prioritizes professional-grade audio quality, producing all output in 48kHz stereo with SynthID audio watermarking for content identification. Its strength is creating coherent, professional-sounding instrumental compositions across many genres, but it does not generate vocals. This makes it well-suited for scoring films, creating ambient atmospheres, and musical exploration.