AI music model
Lyria 3 Pro AI music generation online
Lyria 3 Pro is Google DeepMind’s flagship music model for complete songs — not short loops. In Veontra you turn a prompt (and optionally a reference image) into a finished stereo track with structure, vocals or instrumental, and downloadable MP3.
What Lyria 3 Pro can do
Lyria 3 Pro is designed for full-song generation: intro, verse, chorus, and bridge feel planned rather than stitched from random loops. That makes it a strong default when you want radio-style structure and polished stereo audio up to about three minutes.
You can guide the track with text alone, or add an image reference so mood and visual atmosphere influence the music. Choose instrumental, auto lyrics, or your own custom lyrics. Vocals can be male, female, or both depending on your settings.
In Veontra you also set genre, mood, tempo, and duration before you generate. Optional album covers are available as a separate image step. Lyria outputs are SynthID-watermarked for AI audio provenance.
Billing is a flat credit cost per generation (not per second), so a three-minute song uses the same music charge as a one-minute song — cover credits are added only if you enable cover generation.
How to generate with Lyria 3 Pro in Veontra
- Open the music generator and select Lyria 3 Pro in the model picker.
- Write a clear prompt, then set genre, mood, vocals, tempo, and duration (60–180 seconds).
- Pick lyrics mode — instrumental, auto, or custom — and optionally attach a reference image.
- Generate, preview in your workspace, then download the MP3 (and cover ZIP if you created one).
Who Lyria 3 Pro is for
Creators who need finished song demos for social, trailers, or pitch decks benefit from Lyria’s full-arrangement style. It is a fit when a 15-second loop is not enough and you want a track that feels intentional from start to finish.
Product and marketing teams use Lyria for branded beds with optional vocals, while game and film prototypers use it for cinematic cues before committing to a full score.
If you already have a visual moodboard, the image → MP3 path helps keep the music aligned with the look of a campaign or scene.
How Lyria 3 Pro compares to other Veontra models
Choose Lyria when you want Google DeepMind quality, optional image mood reference, and explicit duration control up to three minutes. Prefer reference-audio steering? Try MiniMax Music 2.0. Need lyrics-first vocals without a reference file? MiniMax Music 2.6. Need instrumentals longer than three minutes? Sonilo v1.1 goes up to about ten minutes.
All four models live in the same Veontra generator and credit wallet — switch models without leaving your workspace.
Lyria 3 Pro FAQ
How long can a Lyria 3 Pro track be?
In Veontra you can request about 60 to 180 seconds. The model targets complete song structure within that window.
Does Lyria support my own lyrics?
Yes. Use custom lyrics mode to paste your words, auto mode to let the model write lyrics from your prompt, or instrumental for no vocals.
Can I use an image as a reference?
Yes. Lyria is the Veontra model with optional image mood reference — attach a picture to steer atmosphere alongside your text prompt.
How is Lyria billed?
Music is a flat credit cost per generation. Enabling an album cover adds image generation credits separately.
Is the audio watermarked?
Lyria outputs include SynthID watermarking for AI provenance, as provided by Google’s music generation stack.
One AI workspace for your team
Extract documents, transcribe audio, and generate images from the same dashboard — with human review before anything is exported.