Crazy stuff with AI going on – seems like a waste of time to write this down even, because at the rate things are advancing, it won’t be interesting in another week or two due to the rapid rate of advancement.

Regardless here are some pics that were created by a simple text prompt in stable diffusion:

There is also AI for speech recognition – “whisper.ai”.

I uploaded the recording of a cassete recording of my great grandfather in a heavy Polish accent and it was able to transcribe fairly accuratelydespite all these challenges. The mistakes it made I had to listen to several times in order to make a better guess.

I also uploaded Hebrew recording with loud background noise and it was able to recognize that it was Hebrew and transcribe fairly accurately.

Pretty amazing when you take into consideration that both the sound and image models are no longer connected to the data that they trained on – they just “know” what you are asking of them…