107: Filling our heads with knowledge
[During] I generated a song about goats, and we illuminated roles in the Persuasion Industrial Complex
[Monday]
Can you hear me?
Much like generative video, I think the challenge with generative audio is going to be use cases. And maybe this is just my crusty Gen X attitude missing the point. But if you’re not routinely creating, editing and working with recordings and music already…I’m not sure generative audio is going to be terribly attractive for routine use.
But the space is tantalizing if you are so inclined. Need an existing voiceover dubbed to another language? A crappy recording improved (Abode’s free enhancer tool is magic)? Or you don’t want to use your own voice to narrate whatever? The tools are there, relatively easy to use, and many are free.
We began the AI for Artists and Entrepreneurs class with an overview of 11 text-to-audio platforms. They are:
Sonantic (now owned by Spotify)
*Both Amazon's and MSFT's tools require coding to use.
My personal favorite is Eleven Labs. So we took a deep dive into Eleven Labs with Alec Wilcock. The Eleven Labs documentation is quite robust and useful.
Then we shifted into music generation, starting with a salient interview hosted by Cleo Abram featuring Grimes and Spotify's CEO Daniel Ek. Key point: While the music industry has lots of experience defining copyright, interpolation, sampling, et al—the entire system is built around compensation for output. Generative AI is all about input. And there isn’t a financial model for that. Yet.
We used this review of 10 AI music generators to level set the space we’re in.
Personally, I’ve found Suno the easiest to use. Here’s a song I made about goats (the prompt: “A jaunty sea-faring hoedown jingle with goats”). They even generated album artwork. And lyrics!
👋🏽 I didn't mention these items in class, mostly because they’re still prototypes, but 1️⃣ Adobe is suggesting its own generative audio platform—above and beyond their current audio editing tools. Here’s the story in The Verge, and a video from Adobe’s Research team. 2️⃣ Google’s MusicLM is an early exploration of text-to-music (via The Verge). 3️⃣ And Meta’s got its own text-to-music generator project called Audio Craft (via The Verge). Meta’s also been testing Voicebox, a text-to-speech generator.
[Tuesday]
Welcome to the Persuasion Industrial Complex
The Persuasion & Marketing class has segued from history, mass media, behavioral economics and politics into the more pragmatic realm of jobs, roles and opportunities. We explored the many definitions of an agency—from the traditional advertising model (as told by the fabulous Ashley Rutstein of StuffAboutAdvertising—a great follow on TikTok or Insta), to public relations, to many of the same roles in the context of corporate marketing. We’ll get into Design, Social, Digital, et al soon enough.
There are so many fascinating nooks and crannies in which you could work across the persuasion arts and sciences these days—and let’s not forget this includes working at Persuasion Arts & Sciences. (Genius name, Dion.)
Then we found our way into a thorny question:
How would we know if (or how or why) persuasion was worth it?
Efficacy has reared its demanding head. So we took some time to revel with Neil Perkin (who writes one of my favorite newsletters) and his guest Dom Boyd, MD of Kantar UK in this video. While it’s loaded with terminology, we’re left with many useful insights. Including Boyd’s take on four methods of measuring persuasion:
o5Relevance: i.e. the persuasion aligns with an audience’s interests and needs. Measured by: Engagement metrics like click-through rates.
Meaningfulness: i.e. the persuasion resonates emotionally or intellectually. Measured by: Surveys assessing brand perception.
Salience: i.e. the persuasion stands out in memory. Measured by: Brand recall surveys.
Difference: i.e. the persuasion is unique from competitors. Measured by: Competitive analysis/performance.
There’s much more to discover and learn, of course.
But next week MCAD is on Spring Break. I’ll be posting at least once, likely mid week.
When we return, the AI course is going to begin putting all the pieces together and focusing on situations and roles where various combinations of generative AI make sense.
And our Persuasion course is going to hit the road and begin meeting practitioners of the dark arts.