Google Lyria 3 Pro
Google DeepMind's latent-diffusion music model, reached via OpenRouter, as a selectable song-generation engine in AutorunX.
What Lyria 3 does
Lyria 3 generates a complete, high-fidelity track — vocals and instrumentation together — from a text prompt or lyric sheet, spanning genres from pop to funk to Motown-style arrangements per Google's own model description.
On AutorunX it's exposed as a prompt/lyrics-to-song engine inside Song Builder: describe the style or paste your lyrics, and Lyria 3 returns a produced track built to that description.
Capabilities
Key features
- Latent-diffusion architecture
- Lyria 3 refines audio latents iteratively rather than predicting tokens sequentially, a technically distinct approach from autoregressive music generators like Suno.
- Trained with musician input
- Google DeepMind describes tuning Lyria 3 with input from producers and musicians specifically to capture musicality — rhythm, arrangement, and flow from note to note.
- Multi-genre, multi-language vocals
- Per DeepMind's model description, Lyria 3 spans genres from pop to funk to Motown and can generate vocals in different languages.
- SynthID watermarking
- Google embeds an inaudible SynthID watermark directly into generated audio, which DeepMind states remains detectable even after compression, slowdown, or re-recording.
- Routed via OpenRouter
- AutorunX reaches Lyria 3 through OpenRouter rather than a direct integration, and it sits in the Song Builder picker alongside Suno v5, Musicful, and ACE-Step.
How Lyria 3 works
Lyria 3 uses latent diffusion applied to temporal audio latents, trained on audio data annotated with text captions at multiple levels of detail. Instead of predicting discrete speech-style tokens one at a time, the model iteratively refines a full audio latent toward the target described by the prompt — a fundamentally different generation process than autoregressive, token-based music models.
On AutorunX, Lyria 3 is reached indirectly through OpenRouter rather than a direct provider integration, which adds a routing layer relative to engines AutorunX connects to natively — one reason it's presented as a selectable alternative in Song Builder rather than the primary path.
It takes the same input as AutorunX's other song engines — a style prompt or a pasted lyric sheet — and returns a finished track with composition and vocals handled together, so switching to Lyria 3 for a given song idea is simply a different choice in the engine picker.
What people use Lyria 3 for
Alternate melodic take
Run a song idea already tried on Suno v5 through Lyria 3 for a distinctly different production and melodic approach.
Multi-language vocal tracks
Use Lyria 3 where its stated multi-language vocal range is useful for a project's audience.
Cross-engine comparison
Generate the same lyric sheet across Lyria 3, Suno v5, and Musicful to compare arrangement styles before finalizing a track.
Provenance-conscious projects
Choose Lyria 3 for content where a built-in SynthID watermark on the generated audio is a wanted signal.
How to use Lyria 3 on AutorunX
Lyria 3 is available in Song Builder, in the Music module.
- 01
Open the service
Go to Music and open the Song Builder editor.
- 02
Set up your input
Enter a style prompt or paste a full lyric sheet for the song you want.
- 03
Pick Lyria 3
Open the engine picker and select Lyria 3 — it's an alternative, not the default, so choose it explicitly.
- 04
Generate
Run the job and review the finished track.
- Billed per completed song, from your shared AutorunX credit wallet.
- 86 credits / song
Tips for better results
- Compare against Suno v5 directly
- Try the exact same prompt on both engines to see how differently a diffusion-based model arranges a song versus Suno's approach.
- Weigh cost against the other alternatives
- At 86 credits/song, it sits between ACE-Step's 44 and Musicful's 122 — a mid-tier alternative worth trying when budget allows.
- Write descriptive, layered prompts
- Diffusion-based models respond well to detailed mood and instrumentation descriptions rather than short keyword lists.
- Paste full lyrics when vocals matter
- Lyria 3 composes melody to fit given lyrics, so pasting a complete lyric sheet gives more control over the vocal content than a style prompt alone.
Lyria 3 — frequently asked
Related models
Suno v5
SunoSuno's song-generation model — full arranged tracks with vocals from a prompt.
Musicful
MusicfulA cloud prompt-to-song API that turns text prompts or lyrics into fully produced tracks, selectable in Song Builder.
ACE-Step
AutorunX ownedAutorunX's owned, self-hosted open-source music engine — Apache 2.0 licensed and the cheapest song-generation option available.
Ready to create with Lyria 3?
Open Song Builder and generate your first music in a couple of minutes.
Free plan, no card required.