Stability AI’s audio generator can now crank out 3 minute ‘songs’

Stability AI , an improved version of the music creation platform. This system allows users to generate up to three minutes of audio via a text request. This is equal to the length of the actual song, so it will also produce the intro, full chord progression, and outro.

First, the good news. Three minutes is big. The previous version of the program reached a maximum of 90 seconds. Just imagine a fake birthday song you could make in the style of a Rob Thomas/Santana track. Another blessing? The tool is free and publicly available on the company’s website, so give it a go.

It works primarily through a text request, but there is an option to upload an audio clip. The system will analyze the clip and produce something similar. All uploaded audios must be copyright free, so this is not meant to imitate what already exists. Conversely, it can be useful for humming a drum part, for example, or stretching a 20-second clip into something longer.

Now the bad news. It’s still AI-generated music. Great as a conversation piece and emblem of a possible future, it’s great for tinkerers and bad for musicians, but that’s about it. Songs can sound great at first until the seams start to show. Then things get a little creepy.

For example, the system likes to add vocals, but not in any known human language. I assume that AI-generated images are in whatever language makes up the text. The vocals sound kind of like real people, and other times they sound like Gregorian chants filtered through space. It is a straight frost in the middle of an unusual valley. The Verge Comparing them to whale sounds, “spiritless and strange”.

Stable Audio 2.0 makes the same weird little mistakes that all these systems do, regardless of output type. Parts can disappear into thin air, replaced by something else. Sometimes there will be melodic elements double out of nowherelike an audio version of extra fingers in AI-generated images.

And there’s the boredom of it all. It’s music in name only. What’s the point if there’s no human connection? I listen to music to get inside the head of another person or a group of people. Despite the constant announcements of artificial general intelligence (AGI), there is no head to head in here. only months left.

So this technology is a must-have for those who make silly birthday videos or bank music. For everyone? Shrug your shoulders. From personal experience, I can say that it is quite fast. The system invented something absolutely terrible Big group song about my cat about a minute.