I've heard that music can be created using AI, but can Gemini do the same? How can I use it? I'm sure many people have questions like that.
On February 19, 2026, Google installed the music generation AI "Lyria 3" in the Gemini app. It is an amazing feature that allows you to create a song with vocals in 30 seconds just by giving instructions via text.
However, there are some cases where it is not possible, such as ``you can only make it for 30 seconds'' or ``you cannot extend it.'' In this article, we have summarized in an easy-to-understand way how to use Gemini's music generation function, what you can do, and what you cannot do.
What is Gemini's music generation function "Lyria 3"?
Lyria 3 is a music generation AI model developed by Google DeepMind. From February 19, 2026, it will be installed in the Gemini app, allowing you to create music in chat format.
Roughly speaking, it is a function that automatically generates a instrumental song with vocals (instruments only) in 30 seconds by simply telling us via text, ``I want this kind of song.'' It also supports Japanese prompts.
The generated music will be automatically embedded with a digital watermark "SynthID" indicating that it is AI-generated content. This is a technology developed by Google that allows you to later determine whether music was created by AI.
How to use Lyria 3 | How to make music in 3 steps
Creating music with Gemini is simpler than you think. Complete with the following 3 steps.
Step 1: Open the Gemini app
Open the Gemini app on your smartphone. A browser version (gemini.google.com) is also available. You must be 18 years or older and have a Google account.
Step 2: Select "Generate Music"
Select "Generate Music" from the tools menu on the chat screen. Alternatively, you can directly enter the prompt ``Make a song that is 〇〇''.
There are 3 ways to make it.
- Text instructions: Tell the genre, mood, and tempo in text, such as "Create 90's style hip hop"
- Choose from templates: 16 genre templates are available, including "Latin Pop" and "Jazz Ballad"
- Upload images/videos: Generates a song that matches the mood of the photo or video
Step 3: Check and download the generated song
If you wait a few tens of seconds, a 30 second song will be generated. You can download it in MP3 or MP4 format and send it to your friends with a shared link. The jacket image will also be automatically created by AI (Nano Banana).
Tips for prompts
In order to create a song that is closer to your ideal, it is effective to include the following elements in the prompt.
- Genre and era: “80s synth pop style” “Modern R&B”
- Tempo: "Up tempo" "Relaxed BPM around 70"
- Instrument: “Acoustic guitar mainly” “Piano and strings”
- Vocal: “Female vocal,” “Low male voice,” “Instrumental (no vocals)”
- Lyrics theme: “The theme of the coming of spring” “The sadness of parting”
You can also specify the lyrics yourself, or if you choose "Leave the lyrics to me," the AI will automatically generate them for you.
What you can't do with Gemini music generation | 5 limitations you should know
Although Lyria 3 is convenient, there are some limitations as of February 2026. Let's check first so you don't end up thinking, "It's not what I expected!"
1. You cannot create songs longer than 30 seconds
This is the biggest limitation. The length of songs that can be generated is up to 30 seconds. It is not possible to create a song with a full chorus. It's good enough for BGM on SNS or as an insert song for short videos, but it's not suitable for full-scale music production.
2. Unable to extend song
Although it was a hot topic in X trends, the function to extend the generated song by saying "make it a little longer" is not supported at this time. Competitive service Suno AI has an Extend feature, but it is not yet implemented in Lyria 3.
3. Unable to modify/edit song after generation
Once a song has been generated, it is not possible to change part of it or replace only the jacket image. If you don't like it, you'll have to change the prompt and regenerate it again.
4. It is not possible to completely copy the voice/style of a specific artist
From the perspective of copyright protection, even if you specify the name of a specific artist, it is not possible to reproduce the artist's voice or song exactly. The instruction to "create an XX-like atmosphere" is reflected as an inspiration, but the design is not a complete copy.
5. There is a limit to the number of times it can be generated
Although it can be used with the free plan, there is a limit to the number of times it can be generated. Once the limit is reached, you will be notified and you will need to wait for it to be reset or upgrade to a paid plan such as Google AI Plus / Pro / Ultra (Google One AI Premium will lower the limit).
Can I use it for free? Differences by plan
As of February 2026, Lyria 3 is available for free if you are 18 years or older and have a Google Account. However, the number of times you can use it varies depending on the plan.
| Plan | Price | Music generation |
|---|---|---|
| Free plan | 0 yen | Available (limited number of times) |
| Google AI Plus | About 1,400 yen per month | More number of generation |
| Google AI Pro / Ultra | From 2,900 yen per month | More generation times |
We recommend trying it out for free first, and if you're not satisfied with it, consider a paid plan.
What happens to the copyright of songs created with Lyria 3?
What I'm curious about is, "Will the songs created by AI become my own?"
According to Google's official blog, songs generated with Lyria 3 are available to users. You can download and share freely.
However, you need to be careful about commercial use (use for the purpose of making money). AI-generated content has limited protection under copyright law, and may not be granted exclusive rights.
In other words, There is no problem for personal use or as background music for SNS, but if you use it for commercial music production, you need to understand the legal risks. All generated songs have a SynthID digital watermark, so you can always identify them as AI-generated.
FAQ
Can Gemini's music generation be used on smartphones?
Yes, it can be used with the Gemini app for Android and iOS. The browser version (gemini.google.com) is also OK. You can use it by logging in with a Google account that is 18 years or older.
Can you make songs with Japanese lyrics?
Yes, Lyria 3 supports 8 languages including Japanese. You can generate a song with Japanese vocals using a prompt like ``Create a spring-themed song with Japanese lyrics.''
Can I upload the songs I made to YouTube or TikTok?
As long as it is for personal use, it can be used as background music for SNS or YouTube. However, commercial use for monetization poses copyright risks, so we recommend checking the terms of use.
What is the difference between Suno and Udio?
Suno AI and Udio can generate songs of up to 4 minutes, and they also have an Extend function. On the other hand, Lyria 3 is limited to 30 seconds, but its strengths are that it can be easily used within the Gemini app and can also be generated from images and videos.
What should I do if I don't like the generated song?
Let's rewrite the prompt to be more specific and regenerate it. The more detailed you specify genre, tempo, instrument, vocal style, etc., the easier it will be to create a song that closely matches your image.
References
- Music generation with Gemini — Google Japan Blog, February 19, 2026
- Lyria 3 - Gemini AI Music & Song Generator — Google, 2026
- Gemini got a song. Limitations discovered using Google Lyria 3, which allows you to create Japanese songs through chat — Techno Edge, February 19, 2026
- Music generation AI "Lyria 3" added to Gemini app, 30-second songs with "Nano Banana" jacket — ITmedia, February 19, 2026
