To not be outdone by Google, Meta has launched its personal AI-powered music generator — and, not like Google, made it out there in open supply.
Known as MusicGen, Meta’s music-generating device — a demo of which will be discovered right here — can flip a textual content description (e.g. “An ’80s driving pop tune with heavy drums and synth pads within the background”) into about 12 seconds of audio, give or take. MusicGen can optionally be “steered” with reference audio, like an current tune, by which case it’ll attempt to comply with each the outline and melody.
Meta says that MusicGen was educated on 20,000 hours of music, together with 10,000 “high-quality” licensed music tracks and 390,000 instrument-only tracks from ShutterStock and Pond5, a big inventory media library. The corporate hasn’t offered the code it used to coach the mannequin, however it has made out there pretrained fashions that anybody with the best {hardware} — mainly a GPU with round 16GB of reminiscence — can run.
So how does MusicGen carry out? Nicely, I’d say — although actually not nicely sufficient to place human musicians out of a job. Its songs are moderately melodic, a minimum of for primary prompts like “ambient chiptunes music,” and — to my ears — on par (if not barely higher) with the outcomes from Google’s AI music generator, MusicLM. However they received’t win any awards.
Right here’s the output from MusicGen for “jazzy elevator music”:
And right here’s MusicLM’s take:
Subsequent, I gave a extra difficult immediate to try to throw MusicGen for a loop: “Lo-fi gradual BPM electro chill with natural samples.” MusicGen surprisingly outshined MusicLM by way of musical coherence, producing one thing that’d simply discover a residence on Lofi Woman.
Right here’s MusicGen’s pattern:
And right here’s MusicLM’s:
To change issues up a bit, I attempted utilizing each instruments to generate a piano ditty within the type of George Gershwin. I say “tried” as a result of, in a effort to forestall the copyright points round generative music instruments, Google carried out a filter within the public model of MusicLM that blocks prompts mentioning particular artists.
MusicGen has no such filter. However the outcomes for “Background piano music within the type of Gershwin,” left one thing to be desired, I have to say:
Generative music is enhancing, clearly (see Riffusion, Dance Diffusion and OpenAI’s Jukebox). However main moral and authorized points have but to be ironed out. AI like MusicGen “learns” from current music to supply comparable results, a reality with which not all artists — or generative AI customers — are snug.
More and more, home made tracks that use generative AI to conjure acquainted sounds that may be handed off as genuine, or a minimum of shut sufficient, have been going viral. Music labels have been fast to flag them to streaming companions, citing mental property issues — and so they’ve typically been victorious. However there’s nonetheless a scarcity of readability on whether or not “deepfake” music violates the copyright of artists, labels and different rights holders.
It won’t be lengthy earlier than there’s steering on the matter. A number of lawsuits making their means via the courts will seemingly have a bearing on music-generating AI, together with one pertaining to the rights of artists whose work is used to coach AI methods with out their data or consent.
For its half, Meta, which isn’t imposing restrictions on how MusicGen can be utilized, says that each one the music MusicGen was educated on was “lined by authorized agreements with the best holders,” together with a take care of Shutterstock.