"Lenny" is only 21 short recordings, stitched together into a crude sequence. It's a remarkably unsophisticated trick on a technical level. It wouldn't be difficult to produce dozens or hundreds of Lennies on quite a small budget. Tweak the script a bit, get someone on Fiverr to record it and you've got a new and unrecognisable Lenny MkII.
Maybe, rather than investing more into Lenny, create some more personas (voice, gender, etc) to add to the variety between calls? Maybe also have an option to hand a call over to the next persona in lengthy calls (so that the caller has to start all over again, in order to explain the issue to the "room mate", "brother", "cousin", etc)...
"Oh, yes ... Oh, so late ... so late, already ... Listen, you have been so friendly with me, but I really have to feed the ducklings. It's well ... beyond their hour ... May I hand you over to my wife, Gertrude? She is such a nice person and knows these things better than me ..."
> It wouldn't be difficult to produce dozens or hundreds of Lennies on quite a small budget.
I'm not sure. Even though it is, as you say, remarkably unsophisticated on a technical level, it is extremely accomplished on a artistic level.
When I first learned about Lenny, I listened to most of his conversations, and it's impressive to hear how often he actually seems to be answering questions or gets distracted in just the right way at just the right time.
The writing (and acting) is just about perfect. I don't know if this is because of sheer genius, or simply lightning in a bottle, but either way I doubt it could be simply reproduced hundreds of times.
Lenny being 'just' a bunch of unchanging messages in unchanging order made it all the funnier and more amazing to me (and it was completely brilliant before I realised).
I remember first seeing it a while ago, posted here, ironically (or perhaps naturally) in a thread about Spamnasty (also excellent).
I'd love to see it as an app with a blacklist for spam numbers or telemarketers. Use a better text generation engine like a modern deep learning based generative model, and deep voice or crowd-sourced recordings.
