I honestly can't think of a reason why I would ever listen to music generated by a computer.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apple Acquires AI Music Startup That Can Generate Dynamic Soundtracks
- Thread starter MacRumors
- Start date
- Sort by reaction score
Now if my HomePod could play a jingle whenever I enter my apartment, and tailor it to me mood, I could live my sitcom fantasy
As a video editor, if it could take care of the temp music instead of me, it could allow me to actually work on the video editing and not on finding the exact perfect temp music for a scene that production will fall in love with but won’t have the budget to buy the rights to and have to replace anyway. I’m all for that.
As a video editor, if it could take care of the temp music instead of me, it could allow me to actually work on the video editing and not on finding the exact perfect temp music for a scene that production will fall in love with but won’t have the budget to buy the rights to and have to replace anyway. I’m all for that.
Anybody here remember Apple's late-2013 SnappyCam acquisition ?
Apple did nothing with the Technology OR the Application, & has since been Leapfrogged in both !
As such, NO guarantee Apple does anything with this OR really ANY acquisition they make.
Apple "typically" acquires small companies for the talents & skills of the founders & key participants, NOT for anything they've already developed.
It's for something Apple has planned that's similar, OR at least in the same ballpark.
Apple did nothing with the Technology OR the Application, & has since been Leapfrogged in both !
As such, NO guarantee Apple does anything with this OR really ANY acquisition they make.
Apple "typically" acquires small companies for the talents & skills of the founders & key participants, NOT for anything they've already developed.
It's for something Apple has planned that's similar, OR at least in the same ballpark.
We’ll see about that. Myself and 1000 of my musician buddies are going to camp downtown, block traffic and loudly play our guitars all day until Apple backs off.
I have friends who are musicians. REAL musicians. I have a relative that is a professional composer for TV, film, video games and more. This kind of s*** will eventually put them out of work. Someday the music in films, the radio, and TV shows will suck, because some suit somewhere will have decided that $10 for an AI soundtrack is cheaper than a real person with talent.
F*** this.
F*** this.
I don't think you get what this is. It doesn't create music using AI, it creates playlists of existing music using AI.
My reaction was more along the lines of:
1) it being used in a next generation version of Cinematic video mode editing or some form of iOS video editor to make YouTube style content creation (including adding DMCA-free music) very easy to do on an iPhone.
2) Upgraded Photos app "moments" compilations with dynamic music that reacts to the edit rather than being relatively disconnected to what's being shown on the display.
I am so excited about this, and I think there are many use cases for this. Using it for Apple Fitness+ to match beats to the user's heartbeat is certainly a possibility I would be excited about. I also think this would be great for Photos and iMovie to create a dynamic soundtrack to films and memory albums. However, I think a big thing that could be enabled by this technology would be automatic DJ mixes in Apple Music based on user playlists. I think this could replace autoplay in its current state and would be a HUGE selling point for AM over competitors. Bookmark me!
Oh, I’d hope not.
If they were working on tech that plays “the right music” in those obtrusive ads (like on AppleInsider), that’s a lack of understanding by the advertiser that I wanted to hear ANYTHING at all in their ads. To be far, I’m sure there are some people that are like “What a surprising treat to hear music when I didn’t expect it, and just my vibe! I shall purchase products now!” or else they wouldn’t have been working on it.The core technology in the “Infinite Music Engine” can be applied in many (likely patented) ways. I understand why there was a focus in the reporting on the key being that it changes due to real-time user interactions (that aligns to things many folks understand, changing tempo theme of a workout, changing the soundtrack in a platforming game), but I think the stronger case is some of those listed higher on the page… like where FCP analyzes your clips, hands that data off to the Infinite Music Engine and have it generate a soundtrack based on that. The user is still interacting with it, but more in a “I’m choosing what you do” way rather than a “take my telemetry as input and create something from that” way. The SceneKit idea is a little of both where the developer is doing “a” and the user experiences “b”.
This is cool, though, I don’t really see how it would be applied in the real world.
If I am working out, will it change the music mid-song just because my heart rate changed? It seems like it would have a 3 to 4 min lag.
Will play automatically based on what I am doing, or do I have to tell it what the activity is? What happens if my heart rate goes too high and I stop working out? Will it play faster music because it thinks I need a lift or will it play slower music because it thinks I need a break?
If I am working out, will it change the music mid-song just because my heart rate changed? It seems like it would have a 3 to 4 min lag.
Will play automatically based on what I am doing, or do I have to tell it what the activity is? What happens if my heart rate goes too high and I stop working out? Will it play faster music because it thinks I need a lift or will it play slower music because it thinks I need a break?
SnappyCam was taking full resolution photos 20-30 frames a second when Apple’s own app was unable to (and no one else was doing it). One can assume that Apple went on to implement some of his algorithms (which may even have been novel enough for a few patents) at the OS level. I wouldn’t be surprised if the camera firmware uses some of his tech.
Apple’s Burst Mode could also be based on this. SO, I doubt it’s true that they did “nothing” with the technology.
Anyone that would use an AI soundtrack is already using “royalty free libraries”, so that’s no loss. The benefit to the listener is that you might actually hear original notes in that low budget score instead of the same 3/4 tunes over and over.
Some and/or All of the above. There’s really nothing you’ve entered that’s out of the realm of “dynamic soundtrack”-ing. If you’re using an Apple Watch the pulse rate change is quicker than 3-4 minutes. Not instant because I think it tracks over time but I can see how there wouldn’t be much delay.
The Apple Watch also has “sensor matching” data so that it can tell if you’re doing one of the few types of “fitness activities” in it’s library. These DO take a bit to kick in, so I’d imagine it’d always be a good idea to select at the start of your workout. It’s all up to how Apple decides to implement it as a feature.
I don’t see the delay as being caused by the heart rate collection, but rather by the length of time left in the current song. Cause it’s one thing to que up a faster song, but if the music changes every time your heart rate changes then the experience will get old quick.
But, if it waits 3 minutes to start the next, faster song the effect of HR on tempo/beat/rhythm is going to be lost.
Last edited:
I see the play here. Have any of you ever owned a brick and mortar retail business? If you have, you've surely dealt with the ASCAP people nosing around. If you have anything other than the radio playing, you have to license it. It's expensive and they will force you into one of the Musak-like business streaming services.
Now imagine if you can generate music for your business with AI. The algorithm knows the vibe of your store, maybe even the current foot traffic level and time of day and can adjust accordingly. It's royalty-free, unique, adaptive, and cheap to make.
Now imagine if you can generate music for your business with AI. The algorithm knows the vibe of your store, maybe even the current foot traffic level and time of day and can adjust accordingly. It's royalty-free, unique, adaptive, and cheap to make.
Why cannot this be considered as bolstering Apple Music to better compete with Spotify and their music algorithms?!
SnappyCam was based-upon an ARM NEON implementation, novel for its day.
It captured 8 Mpx photos @ 20-30 fps, novel for its day.
Apple's Burst mode is strictly a CPU-based implementation, which does NOT use NEON.
Very easy to prove via Benchmarks.
The key is the “dynamic soundtrack”. This isn’t tied to any real piece of music of any particular length. The music is being generated in real time as you go. If you’ve played any recent video games, this is something that’s pumped up or cooled down within a few seconds. A tune with trumpets, tympanis, etc. drumming the tune of your eminent demise and, after you’ve dispatched the monsters, it smoothly transitions back to the more pastoral tune that plays in the background during normal travel.
This is all eminently paramaterizable, though, down to the user. One user may only slow down when they’re done with the workout and they’re in cooldown. Another may have a few high intensity activities in succession with breather’s in between. If the first wants the music to change over a longer period of perceived lower activity and the second wants a quick shift right before pumping back up, that’s all up to the parameters programmed into it.
I didn’t read that at all. I read it as they had an archive of already made songs that it selected from. So it takes music that no one has heard before and then blends it together into an incompressible mess?
I think mainly because it’s an AI that generates music algorithmically, not an AI that analyses music and builds a playlist from already existing music.