How to Make Professional Music with Google Flow Music AI

Google has quietly turned its music AI experiments into a powerful new tool called Flow Music. With it, you can turn a simple text prompt into a full song, complete with vocals, lyrics, and professional production—without needing any music theory or DAW experience.

Whether you want background tracks for videos, instrumental playlists, or full vocal songs in multiple languages, Flow Music is designed to handle it.

What Is Google Flow Music?

Flow Music is Google’s dedicated AI music generation platform, built on the Lyria model—the same family of music models that power Gemini’s audio features. It evolved from a tool called Producer AI, which Google acquired and then rebranded and expanded.

Instead of being a general-purpose chatbot, Flow Music is focused entirely on music creation. You describe what you want, and the system handles composition, arrangement, sound design, and even lyrics and vocals.

Some key capabilities include:

• Generating full songs from text prompts
• Supporting multiple genres and languages (including Japanese)
• Creating instrumental-only tracks or full vocal songs
• Editing and remixing existing AI tracks
• Extending song length well beyond the early 30-second limits seen in older models

If you already use Google’s Gemini Plus, Pro, or Ultra plans, Flow Music ties into that subscription and gives you extra credits for music generation.

Getting Started with Flow Music

To use the tool, go to FlowMusic.app. The old Producer AI domain now redirects there, so everything is consolidated under the new name and interface.

On the main screen, you’ll see a chat-style box where you can describe the music you want to create. You don’t need to know musical terminology—Flow Music is designed to work with both simple and detailed prompts.

For example, you can write something like:

• “Celtic rock music with energetic drums and electric guitar”
• “Chill lo-fi hip-hop for studying, no vocals”
• “Epic orchestral soundtrack with choir, cinematic and emotional”

Flow Music will then generate two variations based on your prompt so you can pick the one you like best.

Customizing Your Default Style with Producer Settings

Flow Music includes a powerful “producer” layer—an extra AI that helps interpret your prompts and shape the final track. You can fine-tune this behavior so the system better matches your usual taste.

System Instructions

Under the “Customize Producer” settings, you’ll find an Instructions section. This works similarly to system instructions or custom preferences in tools like Gemini or ChatGPT.

Here you can define your default preferences, such as:

• Typical genres you like (e.g., synthwave, pop, ambient)
• Usual instruments (e.g., acoustic guitar, piano, strings)
• Preferred mood or energy level (e.g., calm, high-energy, dark, uplifting)

Once set, these instructions are taken into account every time you generate a new track, so you don’t have to repeat the same details in every prompt.

Flows: Reusable Prompt Styles

The Flows section lets you create reusable prompt templates, similar to “gems” in Gemini. You can define a detailed style or setup once, give it a short keyword, and then just reference that keyword in future prompts.

For example, you might create a Flow for “YouTube intro rock track” with your preferred tempo, instruments, and mood. Later, simply typing that Flow name tells Flow Music to apply the full template automatically.

Memories

There’s also a Memories feature that lets Flow Music remember your past interactions to influence future generations. If you prefer each track to be independent, you can disable this. If you like the AI to learn your taste over time, you can turn it on.

Choosing Models, Lyrics, and Inputs

Before generating music, you can tweak a few important options.

Model Selection

Currently, Flow Music offers the Lyria 3 Pro model (previously there was also a preview version). Lyria 3 Pro is the main engine responsible for turning your prompt into audio. In the future, more model options may appear, but for now you simply select the available Lyria 3 Pro model.

Ghostwriter: Automatic Lyric Generation

If you want songs with vocals, Flow Music includes a “Ghostwriter” feature. This is an extra AI layer that writes lyrics for your track based on your prompt.

You can choose between:

• Standard – basic lyric generation
• Pro – higher-quality, more coherent lyrics

When enabled, Ghostwriter will generate lyrics and pass them along with your prompt to the music model, so you get a complete song with words and melody.

Producer Mode

The Producer setting controls how much the system “interprets” and enhances your prompt before sending it to the music model. For most users, the Standard option works well. Behind the scenes, this producer layer can refine your prompt and adjust various generation parameters to get a more polished result.

Using Images and Audio as Inputs

Flow Music is not limited to text prompts. You can also:

• Upload an image – similar to Gemini’s multimodal features, Flow Music can use an image as inspiration for the mood or style of the track.
• Upload existing audio – you can import your own music or a draft idea and ask Flow Music to modify, extend, or transform it.

This makes it useful not just for generating songs from scratch, but also for remixing and evolving ideas you already have.

Editing, Remixing, and Extending Your Songs

Once Flow Music generates two variations for your prompt, you can listen to both and pick your favorite. From there, you have several ways to refine the track.

Editing with Natural Language

You can simply type instructions in the chat to change aspects of the song. For example:

• “Please change the vocal to female.”
• “Make the intro more energetic.”
• “Remove vocals and keep only the instrumental.”

Flow Music will then regenerate the relevant parts while keeping the rest of the song intact where possible.

Remix Mode and New Sessions

There’s also a Remix option. When you click Remix and start a new session, Flow Music brings the selected audio into a fresh chat window dedicated to that track. From there, you can:

• Replace specific sections of the audio
• Change the mood or intensity of only the beginning or middle
• Create new variations of the same idea
• Trim or rearrange parts of the song

For example, you might keep the overall structure but make just the intro more energetic, while leaving the rest of the track unchanged.

Extending Track Length

One of the biggest improvements over the early Lyria demos in Gemini is song length. Where early versions were limited to around 30 seconds, Flow Music now lets you extend tracks to several minutes—5 minutes or more in some cases.

This makes it much more practical for:

• Full songs
• Background music for longer videos
• Ambient playlists for work, study, or relaxation

Downloading and Exporting Your Music

When you’re happy with a track, you can download it directly from Flow Music. Available options include standard audio formats, and you can also generate a music video version.

Keep in mind:

• Audio downloads are usually quick and use fewer credits.
• Generating a video from the song takes longer and consumes more credits.

Once downloaded, you can use the tracks in your video projects, personal listening, or as background audio for streams and presentations. If you’re interested in broader workflows for scoring videos with AI, you may also want to check out this guide to professional sound design with AI.

Exploring Advanced Features: Spaces, Projects, and More

Flow Music includes several extra features beyond basic song generation:

• Song list – a library of all the tracks you’ve generated.
• Playlists – organize your songs into themed collections (e.g., “YouTube intros,” “study beats”).
• Projects – group related chats and songs together, similar to project views in other AI tools.

Spaces: Interactive Music Apps

The Spaces section is especially interesting. It’s similar to a “canvas” where you can build small, interactive web apps around music generation. Examples include:

• DJ pads that trigger different AI-generated loops
• Genre recorders or other experimental tools

At the moment, some of these features can be a bit buggy, but they hint at a future where Flow Music can power more complex, interactive music experiences.

Trim Table and Model Training

There’s also a Trim Table feature aimed at more advanced users. It appears to be related to helping refine and train the underlying AI models, so it’s not essential for basic music generation and editing.

How Flow Music Compares to Other AI Music Tools

Flow Music sits in the same general space as other AI music generators, but with deep integration into Google’s Gemini ecosystem and the Lyria model. If you’re exploring the broader AI music landscape, it’s worth comparing it with tools that focus on different strengths, such as voice cloning or local, offline generation.

For example, if you want to run a powerful model locally instead of in the cloud, you might be interested in running ACE 1.5 XL as a free local AI music generator.

Final Thoughts

Flow Music turns a simple idea—“describe the song you want”—into a surprisingly capable production environment. You can:

• Generate full songs from text, images, or audio
• Let AI write lyrics and handle vocals
• Edit specific sections, change voices, and remix ideas
• Extend tracks to several minutes and export them in standard formats

While some parts of the platform are still a bit rough around the edges, the core experience of going from prompt to polished, studio-style music is already very strong—and it’s likely to keep improving as Google develops the Lyria model and Flow Music platform further.