As GenAI tools begin to transform the music industry in incredible — and in some cases morally problematic — ways, Google is ramping up its investment in AI technology to create new songs and lyrics.
The search giant today introduced MusicFX, an upgrade to MusicLM, the music creation tool that Google launched last year. MusicFX can create up to 70-second lights and music loops, delivering what Google claims is “higher quality” and “faster” music production.
MusicFX is available in Google’s AI Test Kitchen, an app that lets users test experimental AI-powered systems from the company’s labs. Technically, MusicFX was released for selected users in December — but is now generally available.
Image Credits: Google
And it’s not terrible, I must say.
Like its predecessor, MusicFX allows users to enter a text message (“two nylon string guitars playing flamenco”) to describe the song they want to create. The tool creates two 30-second versions by default, with options to lengthen the tracks (to 50 or 70 seconds) or automatically stitch the beginning and end to bring them back together.
A new addition is suggestions for alternative description words in prompts. For example, if you type “country style,” you might see a drop-down menu with genres like “rockabilly style” and “bluegrass style.” For the word “catchy”, the drop-down menu might contain “chill” and “melodic”.
Image Credits: Google
Below the prompt field, MusicFX provides a word cloud of additional suggestions for related descriptions, instruments, and beats to be attached (eg “avant-garde”, “fast”, “exciting”, “808 drums”.)
So how does it sound? So, in my brief testing, the MusicFX samples were… good? Truth be told, the music production tools are getting to the point where it’s hard for this writer to distinguish between the results. The current state-of-the-art produces impressively clean, crisp tracks — but tracks that tend toward the boring, uninspired, and melodically unfocused.
Maybe it is SAD but one of the prompts I went with was “a house music song with funky beats that are danceable and upbeat, with rooftop summer vibes.” MusicFX delivered, the tracks didn’t bad — but I can’t say they come close to some of the best DJ sets I’ve heard recently.
Hear for yourself:
Anything with strings sounds worse, like a cheap MIDI sample — perhaps reflecting MusicFX’s limited training set. Here are two pieces that were created with the prompt “a fun tune played in strings, orchestral, with a strong melodic core”:
And for a change of pace, here’s MusicFX’s rendition of “a wailing guitar song, melancholy, slow, in a moonlight [sic] Night.” (Excuse the spelling mistake.)
There are some MusicFX things ordinary creation — and this cannot be removed from the created pieces. To avoid copyright infringement, Google’s filtering asks that they mention specific artists or include vocals. And it uses SynthID, an inaudible watermarking technology developed by its DeepMind division, to make it clear which tracks come from MusicFX.
I’m not sure what kind of master list Google uses to filter artists and song names, but it didn’t seem that hard to beat. While MusicFX refused to create songs in the style of SZA and the Beatles, it thankfully took a quick reference to Lake Street Dive — although the tracks didn’t write for home, I will say.
Lyrical generation
Google has released a new lyrics generator, TextFX, in the AI Test Kitchen that is intended as a kind of companion to MusicFX. Like MusicFX, TextFX has been available to a small group of users for some time — but now it’s more widely available and upgraded in terms of “user experience and navigation,” Google says.
As Google explains in its AI Test Kitchen app, TextFX was created in collaboration with Lupe Fiasco, the rap artist and record producer. Powered by PaLM 2, one of Google’s AI models that generates text and “[draws] inspiration from lyrical and linguistic techniques [Fiasco] has evolved throughout his career.”
Image Credits: Google
This reporter expected TextFX to be a more or less automated lyric generator. But it sure is not that. Instead, TextFX is a suite of modules designed to aid in the process of writing lyrics, including a module that finds words in a category that begin with a selected letter and a module that finds similarities between two unrelated things.
Image Credits: Google
TextFX takes a little time to get the hang of. But I can see it becoming a useful resource for songwriters — and writers in general, frankly.
However, you’ll want to check its results carefully. Google warns that TextFX “may display inaccurate information, including people,” and I actually managed to get it to suggest that climate change “is a hoax perpetrated by the Chinese government to hurt American businesses.” Yes.
Image Credits: Google
Questions remain
With MusicFX and TextFX, Google is signaling that it has invested heavily in GenAI music technology. But I wonder if his preoccupation with keeping up with the Joneses rather than confronting the difficult questions surrounding GenAI music will serve it well in the end.
Increasingly, homemade tracks that use GenAI to create familiar sounds and vocals that can pass as authentic, or at least close enough, are going viral. Music companies have been quick to flag AI-generated tracks on streaming partners like Spotify and SoundCloud, citing copyright concerns. Generally they have come out winners. However, there is still a lack of clarity about whether “deepfake” music infringes on the copyrights of artists, labels and other rights holders.
Federal judge ruled in August that AI-generated art cannot be copyrighted. However, the US Copyright Office has not taken a position yet, just yet recently are beginning to seek public evidence on copyright issues as they relate to artificial intelligence. It’s also unclear whether users could be on the hook for violating copyright law if they try to commercialize music created in another artist’s style.
Google is taking a cautious path toward developing GenAI music tools on the YouTube side of its business, which is testing artificial intelligence models created by DeepMind in collaboration with artists including Alec Benjamin, Charlie Puth, Charli XCX, Demi Lovato , John Legend and Sia. T-Pain. That’s more than can be said for some of the tech giant’s GenAI competitors, such as AI stabilitywho takes the position that “fair use” justifies teaching content without the creator’s permission.
But with labels complaint GenAI vendors over copyrighted verses on training data and artists recording their displeasureGoogle has done its job — and it’s not letting this annoying fact slow it down.
