Google Flow Music • Hacker News Board

Re: Google Flow Music

#51

Post by janalsncm » Sat, Apr 25, 2026, 12:53 AM UTC

numpad0 wrote: post #35

dabinat wrote: post #2

I’m a little confused about the pricing packages. In what scenario would being able to create 600 songs a month (20/day) be too few? I could understand if this was an API that people built products around, but it seems to be geared directly at consumers.

I don't know anything about these AI tools, but it seems to me like, the yield rates of all these AI media generators are exactly in the range of that of lootbox games. Kids "pull" it like slot machines for set prompt, keeping no more than 1% of outputs. The rest is just thrown away, only potentially useful as negative data. So 600 per month total is probably like just couples per month usable.

That’s a huge amount of messing around to get those handful of songs then. If only 1% are good, you’re pulling the lever 100x more than you should need to.

Re: Google Flow Music

#52

Post by lern_too_spel » Sat, Apr 25, 2026, 12:55 AM UTC

twobitshifter wrote: post #48

Ok, someone explain the use case for this? Jingles? Making a song about your friend / sig other? Are people thinking they are going to sell these songs and create an AI artist?

Marketing jingles for video ads.

Re: Google Flow Music

#53

Post by inerte » Sat, Apr 25, 2026, 12:57 AM UTC

I've tried and couldn't make it sound like Angine de Poitrine, it completely ignores the microtones. Sounds more like Polyphia. It does look like AdP is the answer to AI.... or we haven't trained the models with sufficient microtones, likely due to western music influence.

Re: Google Flow Music

#54

Post by mayukh » Sat, Apr 25, 2026, 12:57 AM UTC

Wow, such hate for anything AI. I haven't used it, I don't care to. But I have teenaged kids and I've seen them hang and tinker with tech and ai. They seem to love it.

Sloppy humans create sloppy output. The AI is just an amplifier, it has no motive

Re: Google Flow Music

#55

Post by dinobones » Sat, Apr 25, 2026, 12:59 AM UTC

I've noticed that all of these music generators suffer from something like "mean" collapse (as opposed to mode, you do get variance, but all results are highly centered around similar sounding songs).

The music is all just very average, it sounds like the most average song with the most average chord progression/drum pattern per genre.

I guess that makes sense if these are most likely next audio token predictors... but it'd be cool if there was a way to inject some type of creativity/novelty into these, or at least tune up the temperature.

Everything so far just sounds like stock library music to me.

Re: Google Flow Music

#56

Post by filoleg » Sat, Apr 25, 2026, 1:01 AM UTC

jmull3n wrote: post #46

It can generate something well produced, but it's really bad at applying taste or direction in the way a human does. The workflow feels wrong. it should be closer to a DAW with chat, where the model outputs stems, samples and arrangement parts instead of one finished track. Then you could target a specific sound, section or idea and actually develop it.

I agree with your DAW UX suggestion very much. I think the writing is on the wall, and Suno is doing exactly that with their Suno Studio.

Re: Google Flow Music

#57

Post by wirgil1 » Sat, Apr 25, 2026, 1:01 AM UTC

DiabloD3 wrote: post #13

Why did Google bother? They're a music store, they sell music, both to own, but also renting their vast library out. Google should learn not to shit where they eat.

big tech companies are 50 companies in a trench coat, there isn't some great aligning directive. Feels like some random side project some employees felt like making.

Re: Google Flow Music

#58

Post by wirgil1 » Sat, Apr 25, 2026, 1:06 AM UTC

philringsmuth wrote: post #43

gnopgnip wrote: post #9

Earlier quoted context omitted.

How does that work with using a camera to take photos?

I’m a photographer, I have almost a quarter million photos in my archives. When I take a photo, 90% of it is composition, during which I move around, analyze lighting, background, aperture, shutter speed, exposure and a whole lot of “what do I want to capture here?” The other 10% is editing, which for me involves minor color adjustments, highlights, shadows, cropping, etc. I make all the decisions. AI can generate an…

So the only difference is the amount of decisions and iteration. If someone spends 5 hours iterating with AI vs 5 minutes on a photo, which one has the better claim to being a creative work

Re: Google Flow Music

#59

Post by throwawayk7h » Sat, Apr 25, 2026, 1:08 AM UTC

Google already killed their main music service in 2020. Why should I trust them with another?

Re: Google Flow Music

#60

Post by jdiff » Sat, Apr 25, 2026, 1:11 AM UTC

tredre3 wrote: post #16

DiabloD3 wrote: post #13

Why did Google bother? They're a music store, they sell music, both to own, but also renting their vast library out. Google should learn not to shit where they eat.

Welcome to 2026's reality, most new music is already AI-generated. I don't like it, but it is what it is. YT Music is already full of AI slop, those tools aren't changing that. If anything it gives Google control of the entire production->sale->delivery process. I'm honestly not seeing a downside for Google here, can you elaborate?

Most new music by what definition? I'm certain more stuff is being churned out by these automated tools than genuine human creativity, but that doesn't make it economically relevant if the only use it's seeing is random high school kids' YouTube channels. It's not seeing streams on services, it's not bringing in revenue once created.