Google just updated Lyria 3.5, its AI music tool, so you can now redo a single part of a song instead of starting the whole thing over. If you have ever made a track you almost liked, this is the kind of change that saves you from scrapping it and building again from scratch.
The Gist
- Lyria 3.5 lets you edit one section of a song without regenerating the entire track.
- It can stretch a short melody into a full song and fine-tune the tempo and length of each element.
- It lives inside Google Flow Music and makes tracks from 30 seconds to 3 minutes.
Have ChatGPT Recap This Article
ChatGPTMeet Lyria 3.5, Google’s music generator
Lyria 3.5 is Google’s tool for making music from a text prompt. You type what you want, a mood, a style, a few lyrics, and it produces a track with instruments and vocals. It is aimed at anyone curious about making a song, not just musicians.
The new version rolled out inside Google’s Flow Music app, where the company laid out the update. Google says the model now writes richer melodies, follows your lyric prompts more closely, and produces vocals that sound more natural and better pronounced.
On the flip side, the headline is not the sound quality. It is that you finally get to change a finished track without throwing it all away, which is the part beginners struggle with most.
How editing just one part actually works
The new feature is called Selective Section Painting. In plain terms, you point at one stretch of the song, the chorus, a weak verse, a rough ending, and tell Lyria to redo only that stretch while the rest stays exactly as it was.
It also works the other way. You can hand it a short melody and ask it to grow that clip into a full song, so a ten-second idea hummed into your phone can become a complete track.
On top of that, you get finer control over the pieces. Lyria 3.5 lets you adjust the tempo and length of the vocals, the drums, the bass and other parts on their own. This is the same move we saw when Google opened up free image generation to beginners in the US: take a pro feature and make it clickable for everyone.
Keep learning on AI Noobies:
- ChatGPT Resume: A Beginner’s Step-by-Step Guide
- Amazon Nova AI Models Are Getting Quietly Shelved
- Apple Smart Glasses Delayed Over Privacy Fears
What a “section” and a “stem” really mean
Two words are worth knowing here. A section is just a chunk of the song, like the intro, the verse or the chorus. Editing by section means you fix the chorus without touching the verse.
A stem is one layer of the sound on its own, the vocals alone, or the drums alone. When Lyria lets you tweak the tempo of the bass without changing the singing, it is working stem by stem, the way a real studio does.
Why does this matter to you? Because it turns “generate and pray” into “generate, then fix the one bit you don’t like”, which is how normal editing on a phone already feels.
How to try it, and why it stirs a bigger debate
To try it, you open Google Flow Music, describe the track you want, and let it build something between 30 seconds and 3 minutes. Then you pick the weak section and ask for a redo, or extend a short idea into a full song. That is the whole beginner loop.
Set against this, easier AI music also feeds a louder argument you will keep hearing. Streaming platforms are already trying to flag which songs were made by AI so listeners know what they are hearing, and every new tool that makes tracks this simple pushes that question harder.
For you, the takeaway is small and useful. You can now make a passable song and repair the one part that bugs you, which a year ago meant starting over. The bigger fight over who counts as the artist is one the whole industry is still working out.
Stay tuned on AI Noobies.
About the Author
