And the Grammy for best AI Prompt goes to...
Music & AI
Ever since I can remember (which is a long time ago), I’ve been fascinated by music.? How can a song sound happy or sad, angry or calming, passionate or passive?? How is it that a combination of chords, melodies, lyrics and rhythm can bring about such emotions?? As Cole Porter put it:
“There's no love song finer,
But how strange the change,
From major to minor,
Ev'ry time we say goodbye.”
When I’m working from home, I play the piano or guitar several times every day.? It’s an effective distraction technique to free up the problem-solving part of my brain, allowing it to focus subconsciously on a challenge.? I have several times leapt up from the piano, Eureka style, having resolved an issue that’s been troubling me for some time.? I am far from above average in my competency as a musician; I am prone to missed notes, to erratic timings, to obsessive use of sustain pedals.? But it doesn’t detract from the sheer enjoyment; the sheer pleasure of playing and generating music.??
When I’m actually working at home and not playing an instrument, I listen to what my wife describes as “Angry, difficult listening, dance music” such as Special Request, Blanck Mass and Tim Reaper.? It’s usually very fast, 160bpm + and played ridiculously loudly.? Somehow – I have no idea how – it helps me concentrate and plough through backlogs of work.
I have a song in my head all the time that I am seemingly unable to control, often inspired by words or phrases uttered by people nearby[1].? That the human brain is capable of recreating an entire song – instruments, a melody and lyrics, despite perhaps not remembering all of them – is astounding. We’re replaying a song live in our heads, imperfections and all.? In 2023, scientists were able to capture the electrical activity of several brain regions of epileptic patients relating to tone, rhythm, harmony and lyrics in a song the patients were told to replay in their heads.? Using Machine Learning, the scientists were able to identify and recreate the song – “Another Brick in the Wall” by Pink Floyd (chosen, apparently, for its music complexity).
I don’t see music as some people with synaesthesia do, but I definitely feel it. The brain’s dopamine hit from music is the best drug available[2] (probably).? Other than the birth of my daughter and getting married, the happiest I’ve ever been was watching the US band Jurassic 5 perform live at the UK’s Glastonbury Festival. ?I was transfixed, grinning like an idiot for 40 solid minutes, swept up in hip-hop beats, the rhythm of the 5 MCs and the rumbling bass tones of Tuna Fish’s raps.
I love the passion in Brooke Bentham’s voice when she sings “I love it when you put your legs over my thighs” in her song “I need your body”.? I feel the angst in Bob Smith’s voice when he sings “It doesn’t matter if we all die” in The Cure’s “100 Years”.? I am forever moved by the beauty in Guy Garvey’s voice when he tells his girlfriend “You are the only thing in any room you’re ever in” in Elbow’s “Starlings”.
During the Covid “Lockdown” of 2020, the drum & bass DJ High Contrast performed a live set with his band on YouTube, split over their five separate, socially distanced locations.? He finished the performance with an incredibly moving synchronised video, showing the same band performing a live gig from the previous year in front of thousands of gloriously happy people. ?At the time of watching, my family were in a childcare bubble with friends of ours, and every Friday night my daughter and her friend would dance around the lounge to that 30-minute YouTube clip and for a while the world was all OK again. ?It’s an obviously live performance as the vocals are not always 100% to pitch, whereas in the recorded version of the same song, everything is perfect.
And, with apologies for the delay, this is my point: the humanity of music is embodied in its imperfections (even with studio-based dance music).? It’s the guitar squeaks left in recordings, the slightly out-of-key vocals, the drum beats not fully quantized[3], the unplanned, none-rehearsed additions that a perfectionist may strive to correct.? When the quirks and nuances are removed, they take the passion with it.
In The Breeder’s Cannonball, the bass player Josephine Wigg plays the song’s introduction a semitone flat, only correcting herself when the guitars appear.? But the band decided to leave this “error” in the recording, and now everybody accepts it as part of the song.? ?Appearing roughly 4 seconds into Roxanne by The Police is an atonal piano sound, shortly followed by a laugh.? This is from Sting, the bass player and singer, who accidentally sat on the studio piano.? But again, the band and producers decided to leave it in the recording.?? In U2’s The Unforgettable Fire, the drummer, Larry Mullen Jr. counts off with his sticks during the ambient guitar intro before quietly, yet audibly, exclaiming “Oh shit!” when he realises he wasn’t supposed to start.? All left in the recording ?– and even left untouched in 2009’s remastered version.?
领英推荐
Such imperfections are what makes the music sound human, expressive and full of energy.? Yet over the past twenty years or so, significant progress has been made in perfecting recording and production techniques – such as removing audible breaths from singers, fully and automatically quantizing drums and tuning vocals to 100% pitch centre: perfect music; lifeless music.? Like a Picasso with all asymmetry and distorted figures removed; like an Olympics event where nobody ever drops the baton, performs a mistimed jump or makes an ill-judged early dip for the line.? Perfection removes the expression, the randomness; the humanity.?
Advances in technology have allowed us to reach the point where studios can produce the musical equivalent of Ultra Processed Food: ultra processed music, with repeated chord progressions and common patterns that people recognise, all tweaked and sounding absolutely perfect. But people get addicted to UPF for exactly the same reason – it’s designed to appeal to the part of your brain that seeks instant gratification. But there’s no depth there, and certainly no beauty. ?And UPF is very, very bad for you.
Software tools such as Udio and Sudo can generate entirely new songs based on user prompts, following specific phrases to define styles of genre, lyrics and suggested background story.? Some of these creations are incredibly difficult to distinguish from those produced by humans in a modern recording studio.? Then again, we’ve spent the last 20 years or so trying make music sound like it was made by robots; we seem to have finally succeeded.? ?
I’ve written previously about the pros and cons of AI on humanity, and touched upon the merits of AI as a creative tool.? It has been well-documented that most people would prefer AI to do the washing and vacuum the house, freeing up humans for the creative tasks, rather than the other way around.? Yet so much focus has been on attempting to enhance AI’s pseudo-creative capabilities.? We’ve reached a point where robots and AI can write songs and generate art but can’t load and unload a dishwasher (sidenote: at the time of writing, robots do exist which are capable of loading plates or unloading glasses, but are sadly incapable of handling the full range of items within a dishwasher, particularly cutlery, which kind of makes the whole pursuit pointless).
George Orwell’s 1984, published 75 years ago (1949), was certainly prescient, perhaps prophetic, in its future vision for music: “The tune had been haunting London for weeks past.? It was one of countless similar songs published for the benefit of the proles by a sub-section of the Music Department.? The words of these songs were composed without any human intervention whatever on an instrument known as a versificator.? But the woman sang so tunefully as to turn the dreadful rubbish into an almost pleasant sound.”
I don’t want imperfections in my aeroplane design, but I will always want it in my art.? I love “Release Radar” Fridays on Spotify, when its algorithm curates a list of new music it thinks will appeal to me – some based on bands or artists I’ve previously told it I like, but some new suggestions.? How long until Spotify is generating music it thinks I might like??
[1] This is not always helpful. All it takes is for somebody to ask “Did everyone have a good weekend” and my brain switches over to “Get set for the weekend, I’m showing out” by Mel and Kim.
[2] There are areas of the brain activated by music but little else. This can have a profound, albeit sadly temporary effect on patients with dementia or stroke victims.
[3] Quantizing relates to the movement of notes to align them to a perfect time signature grid.?
Head of Solutions Architecture @Gousto | Product Architecture | Modern Tech Platforms | Bringing people to the problem.
1 天前David Bowie said it: "Music itself is going to become like running water or electricity. So it's like, just take advantage of these last few years because none of this is ever going to happen again. You'd better be prepared for doing a lot of touring because that's really the only unique situation that's going to be left." I believe in human creativity.
Please no !!!!!!!!!! OMG, the day the Music died !!!!!
Solution Architect at Halfords. Passionate about all things AI/ML.
2 个月I believe it's here to stay. With voice cloning, production companies will find new ways of regurgitating old songs and find new depths of the barrel. With a more positive spin, it can be a way to extend the legacy of artists who have been taken too soon. It will also go hand-in-hand with text-to-video, ultimately providing musical scores, jingles, and sound effects for everything from adverts to blockbusters. #DeathToAllButMetal
Portfolio & Programme Management Professional | HEART UK Charity Trustee
2 个月Depends if AI becomes sentient. Then we may be haunted by hearing music inspired by their Pinocchioesque pain, dreams and desires…