Speech Synthesis

[Corpsey]

 

[Corpsey]

 

[WashYourHands]

The Johnson one is good because it nails the rhythm and cadence

 

[Corpsey]

I mostly just find these funny but what I'm intrigued by about it is where its going "wrong", where it doesn't sound like Frank Sinatra singing dancing queen cos the pitch is totally off and the AI is singing in a way a person might never hit on.

 

[Corpsey]

I guess the potential here (when it becomes smoother) is for people to ultimately write new Frank Sinatra songs and have him sing them.

 

[Corpsey]

When I used to make music on reason I sometimes would let the software generate a random melody or rhythm and build on it. And actually of course composing music is to a greater or lesser extent a series of accidents. Accidents you choose to preserve or not. Though much less so for people who know music theory or even are just really good at writing music. For ppl like me it's accidentally hitting on something decent from time to time.

 

[Corpsey]

Maybe there'll be an interesting convergence of the auto tuned vocals you get in rap and rnb and pop now with AI generated melodies.

 

[linebaugh]

the frank sinatra AI sounds hammered

 

[mvuent]

i really like this stuff. muddles the distinction between memes and art even further.

agreed about when it goes wrong. probably has a ton of potential for creative misuse. obviously using it the way it's intended can already be disorienting, but i bet that's what will produce the most interesting and futuristic results. at least at first.