5 prompt tips for Lyria 3 and Lyria 3 Pro
Highlights
AI music generation works best when you know how to guide it. The right prompt can turn a rough idea into a usable track — whether you’re scoring a video, building a soundtrack, or just exploring creative directions.
Lyria 3 is Google’s most advanced music generation model. It can create songs up to three minutes long, complete with instruments, vocals, and lyrics, from both text and image prompts. The model delivers professional-grade audio and offers broader commercial use rights than other AI music models. You can learn more about the model and AI music here.
The key to getting great results is knowing how to prompt it. Here’s how to get the most out of Lyria 3 and Lyria 3 Pro.
1 . Start with the right prompt structure
A simple structure consistently produces strong results. Here is one example:
mood + genre + song subject
This format helps the model understand the emotional tone, musical style, and lyrical theme. Here are a few examples:
- {Mood} Epic + {genre} rock + {song subject} ballad about dogs: “An epic rock ballad about dogs.”
- {Mood} soft + {genre} indie + {song subject} love: “A soft indie song about love.”
Even simple prompts can produce strong results. Adding detail can improve accuracy and creative control.
2. Add details for more control
When you include specific musical details, Lyria has more information to work with. This often leads to more precise results.
Instead of writing: “Rock song.” Try something more descriptive like, “Early 2000’s punk-rock band, lead male vocal singer with gritty voice, heavy distorted guitars, loud drum fills. The chorus should be catchy and anthemic.”
You can guide the model by describing:
- Genre or subgenre
- Era or stylistic influence
- Instrumentation
- Mood and energy
- Vocalist gender or general vocal style
- Song subject or lyrical theme
The clearer your description, the easier it is for the model to generate music that matches your intent.
3. Understand what Lyria follows well
Knowing what the model responds to best helps you write prompts that deliver consistent results.
Excellent prompt adherence
Lyria generally follows these instructions very well:
- General genre
- Vocalist gender and general vocal style
- Mood
- Instrumentation
- Specific BPM
- Energy
- Song subject and theme for lyrics
These elements should be the foundation of your prompt.
What works less reliably
Some instructions may be interpreted more loosely. Precise timestamps are often rounded to the nearest musical bar. This makes musical sense, but it may affect time-sensitive workflows such as scoring or syncing music to video edits.
Certain types of instructions are not currently interpreted well by the model.
These include:
- Musical key
- Very specific instrumentation
- Highly detailed vocal texture descriptions, such as raspy, rugged, or gravelly voices
- Some specific genres, which may be misunderstood
- Multiple genres in one prompt — the model usually selects one
- Pitch references, such as “low octave male voice”
- Audio effects or production instructions, like tape hiss or filtering
- Target volume
For best results, focus on the elements the model understands well — genre, mood, energy, and theme.
4. Don’t use artist or song names
Prompts that reference specific artists or songs will block generation, but there is a workaround to get the song you are imagining.
Instead of writing: “Song in the style of Taylor Swift,” describe the musical characteristics: “A country-pop crossover with acoustic guitar and a catchy melodic hook with storytelling lyrics, bright, airy vocals.”
This gives the model stylistic direction without referencing a specific artist.
5. Speed up prompting with Artlist tools
Artlist includes tools that make it faster to create prompts and lyrics.
Auto Prompt and Auto Lyrics
You can automatically generate a prompt or lyrics based on existing inputs. This helps when you want to explore ideas quickly or refine a starting concept.
Refine results with advanced settings
Artlist also gives you additional controls to guide the output without rewriting your prompt.
You can adjust:
- Genre: Cinematic, electronic, ambient, hip hop, pop, corporate, rock, or funk
- Mood: Uplifting, exciting, powerful, happy, epic, hopeful, peaceful, or dramatic
- Video theme: Commercial, social media, documentary, film and cinema, or wedding
- Tempo: Six options ranging from slow to fast depending on your desired BPM
- Song length: From 30 seconds to 3 minute songs, to fit your project needs
You’re not limited to the dropdown menus. These elements can also be described directly in your prompt.
Artlist Sound
Also found in the AI Music advanced settings, Artlist Sound enhances prompts to improve the quality and style of the generated music. It helps the model produce results that move beyond a generic stock sound while preserving the creative direction of your prompt.
Prompt examples in action
The best way to understand prompting is to hear the results. Here are several prompts used to generate example tracks with Lyria 3. Each one demonstrates a different level of detail and creative direction.
- Detailed genre prompt
This example shows how detailed prompts can shape sound design, rhythm, and overall energy.
Prompt: A futuristic UK dubstep track with deep, wobbling basslines and sharp, syncopated drum patterns. The sound design should be gritty and experimental, featuring distorted growls, metallic textures, and glitchy transitions. Include vocal chops that feel mysterious and atmospheric. The track should have dark, underground vibes with cinematic build-ups leading to heavy, high-energy drops. BPM around 140 – 145, perfect for a festival or warehouse rave in 2025.
- Classic short prompts
A short prompt can also work. Specifying a musical era and form provides enough stylistic direction for the model to generate a recognizable composition.
Prompt: 18th century sonata.wav
This prompt focuses on genre, performer type, and scale. Even minimal instructions can produce a clear musical direction.
Prompt: K-pop girl band, epic pop anthem.
- Orchestral emotional prompts
This combines historical style with cinematic storytelling. Words like “climax” and “inspiring” guide the emotional arc of the music.
Prompt: Romantic symphony, 19th century, climax, inspiring, film score.
The key to better AI music prompts
Great prompts balance clarity and creativity. Start with the core structure — mood, genre, and subject. Then add the musical details that matter most, such as energy, era, instrumentation, and theme.
With clear prompts and the tools available in Artlist, you can generate music that feels intentional, expressive, and ready for real creative work. It’s time to test these tips for your own projects — try AI Music today.
